Recent trends in Income Inequality: Labor, Wealth and More ......11.4 over the same period. And...
Transcript of Recent trends in Income Inequality: Labor, Wealth and More ......11.4 over the same period. And...
Forthcoming: Research in Labor Economics
October 15, 2010
Recent trends in Income Inequality:
Labor, Wealth and More Complete Measures of Income
Timothy M. Smeeding
Institute for Research on Poverty
Robert M. La Follette School of Public Affairs
U. Wisconsin, Madison
and
Jeffrey P. Thompson
Political Economy Research Institute
U. Massachusetts, Amherst
The opinions and conclusions are solely those of the authors and should not be construed as
representing the opinions of our employers. We thank Anthony B Atkinson, Andrea Brandolini,
Gary Burtless, Daniel Feenberg, Robert Haveman, Karen Nordberg, Emanuel Saez, Eugene
Smolensky, Robert Pollack, John Nye, Jodi Sandfort, Barbara Torrey, Rosa García Fernández
and Bruce Meyer for comments and suggestions on a previous draft; and also the participants in
seminars given at the OECD, The Tobin Project, Washington University, St Louis, Syracuse
University, and the MacArthur Network on the Family and the Economy. We thank Kati Foley,
Karen Cimilluca and Mary Santy for data and manuscript preparation.
Please send comments to Tim Smeeding ([email protected]) or Jeff Thompson
2
Abstract
The impact of the ―Great Recession‖ on inequality is unclear. Because the crises in the
housing and stock markets and mass job loss affect incomes across the entire distribution, the
overall impact on inequality is difficult to determine. Early speculation using a variety of narrow
measures of earnings, income and consumption yield contradictory results. In this paper, we
develop new estimates of income inequality based on ‗more complete income‘ (MCI), which
augments standard income measures with those that are accrued from the ownership of wealth.
We use the 1989-2007 Surveys of Consumer Finances, and also construct MCI measures for
2009 based on projections of assets, income, and earnings.
We investigate the level and trend in MCI inequality and compare it to other estimates of
overall and ‗high incomes‘ in the literature. Compared to standard measures of income, MCI
suggests higher levels of inequality and slightly larger increases in inequality over time. Several
MCI-based inequality measures peaked in 2007 at their highest levels in twenty years. The
combined impact of the Great Recession on the housing, stock, and labor markets after 2007 has
reduced some measures of income inequality at the top of the MCI distribution. Despite
declining from the 2007 peak, however, inequality remains as high as levels experienced earlier
in the decade, and much higher than most points over the last twenty years. In the middle of the
income distribution, the declines in income from wealth after 2007 were the result of diminished
value of residential real estate; at the top of the distribution declines in the value of business
assets had the greatest impact.
We also assess the level and trend in the functional distribution of income between
capital and labor, and find a rising share of income accruing to real capital or wealth from 1989
to 2007. The recent economic crisis has diminished the capital share back to levels from 2004.
Contrary to the findings of other researchers, we find that the labor share of income among high-
income groups declined between 1992 and 2007.
3
I. Introduction
This paper is an attempt to capture the effects of secular and cyclical forces on the
inequality of income across Americans who are suffering through the ―Great Recession‖ and the
period of slow employment growth and housing market stagnation that has followed. A full
accounting of inequality in this period will have to wait years, as impacts of the recession and its
aftermath are still unfolding, and the necessary data will not be available until 2011. The most
current micro-data that can be used to analyze income distribution are from calendar year (CY)
2009 (Current Population Survey [CPS] income or poverty), or CY 2007 (Survey of Consumer
Finances [SCF] wealth).
Based on currently available data, however, we do know quite a lot about some of the
economic hardships resulting from the recession. The economy lost jobs every month between
December 2007 and October 2009 – four months after the official end of the Great Recession –
8.3 million jobs in all, and unemployment rose from 5.0 to 10.1 percent (NBER, 2010). The
incidence of job loss has been particularly severe among young workers, and those with lower
levels of education Total employment declined by less than five percent, but among teens it has
fallen 20 percent and among those with high school degrees or less it has declined 7 percent
(Engemann and Wall, 2009). Poverty rose in 2009, and forecasts based on available
employment and food stamp data indicate it will likely go up again in 2010 (Census, 2010;
Monea and Sawhill, 2009).
Expected changes in the distribution of income in 2009, 2010, and beyond, though, are
not as clear. Past recessions (excepting the Great Depression of the 1930s) tended to hurt people
at the bottom of the distribution to a greater extent than people at the top (Atkinson, 2009). These
effects are and were tempered by the safety net, and are driven by the loss of labor market
earnings, which recovers when employment recovers. However, a major aspect of the recent
4
recession has been the drop in property income values, financial assets, and home prices, as well
as employment losses. Because all parts of the income distribution have suffered losses of
income and wealth, the impacts on the overall distribution are more difficult to determine.
Preliminary analysis and speculation over shifts in the distribution suggests a range of
potential outcomes. There is some evidence that the collapse in the stock housing markets have
produced declining CEO pay, lower dividends, and reduced Wall Street bonuses, which could
cause the income gap to shrink ―at the expense of the wealthy‖ (Davis and Frank, 2009;
Leonhardt and Fabrikant, 2009). Looking to data on consumption, some researchers have found
evidence of declining inequality between 2006 and 2009 (Meyer and Sullivan, 2010; Heathcote
et al. 2010a, 2010b, and; Parker and Vissing, 2009). Much of that decline is attributable to a
notable drop in consumption at the top of the distribution, partially reversed in 2009 as the
Obama ARRA plan boosted durables spending and the stock market recovery took hold (Parker
and Vissing, 2009; Petev, Pistaferri, and Saporta, 2010). Overall consumption still fell in 2008
and 2009 combined, but the change in inequality is less certain once we look at the 2009 and
early 2010 data.
Early indicators from some standard income inequality measures from the Census
Bureau, however, suggest that high income shares, as well as Gini and Theil indices, rose
between 2007 and 2009 (Census, 2010). The major losses in income, in proportional terms, were
experienced by the 80th
and 10th
percentiles, with relatively smaller losses for the 90th
percentile
(Smeeding and Thompson, 2010). These findings are fully consistent with those of Krueger et al.
(2010) and Heathcote et al. (2010a, 2010b), who also find earnings and disposable income
inequality rising secularly in rich countries and also in recessions, including this recession
(Heathcote et al., 2010b) and especially for bottom income units. Because of top-coding in the
5
Current Population Survey, though, these data can tell us little about what is going on at the very
top of the distribution.
Data with broad measures of income and that also contain detailed information for
households at the very top of the distribution, are not yet available to give an updated
understanding of inequality. The Congressional Budget Office ―tax burden‖ series, for example,
are only available up through 2007 (CBO, 2010). Similarly, the Survey of Consumer Finances as
well as the IRS tax data used in analysis of high incomes are only available through 2008
(Smeeding and Thompson, 2010; Piketty and Saez, 2006, 2010). But, as Burkhauser, et al (2009)
show - using non-top coded Census Income data - most of the change in income inequality over
the past decade has been amongst the rich. However, even these data exclude the vast majority of
capital income—the issue to which we now turn.
In the remainder of this paper, we will: first, briefly review some of the different
approaches to analyzing trends in income distribution; second; describe our method for
calculating a ―more complete‖ measure of income (MCI), third, compare levels and trends – for
recent years and across the last couple of decades – for inequality using MCI and other standard
income measures, fourth, describe the impact of using MCI on the trends in capital vs. labor
shares, and finally, discuss some potentially policy implications of these trends.
The MCI income concept incorporates a broader range of the resources available to
households than the definition of income in the typical survey, and, as such, is a better
representation of economic ―well-being.‖ Motivated by the classic Haig-Simons income, MCI is
intended to reflect the possibility to consume, and is also arguably a better representation of well
being than actual measured consumption. Estimated with data from the Survey of Consumer
Finances, MCI results in higher income across the distribution, but especially at the top end. We
also find a greater trend toward income concentration at the top of the distribution using MCI
6
than do other analysts. A number of standard measures of inequality using MCI peaked in 2007,
after rising relatively steadily since 1989, including the Gini index, the 99/50 ratio, and the
income shares of top 1 percent and next 4 percent. Nearly all of the increase in inequality is the
result of large gains at the very top of the distribution, with little evidence of rising inequality at
the bottom of the distribution. The Great Recession appears to have halted, temporarily at least,
the trend toward greater inequality. Any declines, however, have so far been modest, leaving
inequality as high as any point before the 2007 peak.
We also assess the level and trend in the functional distribution of income between
capital and labor. We find that properly measured, the labor share is closer to 55 percent of total
income than the 75 percent that is sometimes claimed. The results using MCI suggest that,
contrary to the findings of Piketty and Saez (2003, 2006), the capital share of income at the top
of the income distribution has risen in recent decades (as also found cross nationally by Glynn,
2009). By 2007, income from capital accounted for more than half of MCI among the top few
percentiles of the income distribution.
II. Approaches to Understanding Inequality and the Distribution of Income
For some time there has been widespread concern about growing inequality in the
distribution of household income in the United States. The US Census Bureau shows the Gini
index of household income rose from .40 to .47 between 1967 and 2009, and that the ratio of
incomes of households at the 90th
and 10th
percentiles of the income distribution rose from 9.2 to
11.4 over the same period. And while there is a general consensus among researchers that
income inequality has increased in the United States and much of the rest of the world
(Brandolini and Smeeding, 2009), there is less agreement over how much it has increased, or
whether income is even the most important factor in understanding inequality, let alone the
causes of the increase.
7
Labor economists have shown that inequality in hourly wages increased considerably
over the same period (Autor et al, 2008). With earnings representing the single largest portion of
household income, some argue that trends in earnings inequality are the key factor behind
inequality in the US income distribution.1 A number of recent provocative studies highlight the
role of extremely high earnings among ―superstars,‖ CEOs, athletes, rock stars, and celebrities
(Kaplan and Rauh (2010); Walker (2005), and Gordon and Dew-Becker (2005)), but these
papers are only able to identify about 25-30 percent of even the highest income earners.
And, labor income in the form of wages had declined to 50.2 percent of national income
by the third quarter of 2006 – a 50-year low as a share of national income (Aron-Dine and
Shapiro, 2006; Bureau of Economic Analysis, 2010; Goldfarb and Leonard, 2005). Even after
adding together labor income (even including supplements or employee benefits) and corporate
profits, which peaked at 13.7 percent of total national income in the third quarter of 2006 after
rising for three decades, there is still more than a fifth of the nation‘s economic pie missing.
Other uncounted components of National Income such as net interest, proprietor‘s income and
rental incomes are largely missing from micro date based income distribution calculations (see
Table 1).
Meyer and Sullivan (2010) argue that levels of income inequality are not as great as
suggested by the Census Bureau, and that the emphasis on income itself is misplaced. With
appropriate adjustments for household size, taxes, and transfers, Meyer and Sullivan (2010)
show that the 90/10 ratio was 5.3 in 2008, up from 4.1 in 1979. More important, they argue, is
that consumption is a better proxy for well-being or even permanent income than the income
measures used in most of the inequality research (Also see Slesnick (1994, 2001.))2
1 See Autor, Katz, and Kearney, 2006; Katz, 2006; Lemieux et al., 2007; Lemieux, 2006; Cowen, 2007.
2 Other work by Michelangeli et al (2009) does not use consumption data as a proxy for lifetime income, but instead
develops a method for detecting changes in permanent income concentration when only consumption data are
8
Consumption inequality has showed no trends toward greater inequality in recent decades, and
has – as mentioned above – declined in the last few years.
Consumption is a strong predictor of different measures of hardship (Meyer and Sullivan,
2003), but it is deficient in some important respects as a measure of well-being. As Dickens‘
famous line suggests, it might be better to treat the debt-financed consumption of low-income
households whose consumption far exceeds their income instead as a measure of hardship:
―Annual income twenty pounds, annual expenditure nineteen six, result happiness.
Annual income twenty pounds, annual expenditure twenty pound ought and six,
result misery.‖
- David Copperfield
And by focusing on the 90th
percentile of the distribution, much of the consumption-oriented
research misses what is going on at the very top of distribution.
Several analysts have suggested that most, if not all, of the gains in incomes from rapidly
expansion of productivity in the 1990 and early 2000s accrued to the richest 1-5 percent of
Americans (Gordon and Dew-Becker, 2005; Piketty and Saez, 2003; 2006).3 This result is
supported by the analysis of top-coded Census Income data by Burkhauser, et al (2009). The
long-term analysis by Atkinson, Piketty, and Saez (2009) shows that since the early 1970s
income growth among the top five percent (particularly the top one percent) has far outpaced the
rest of the nation.
Even in micro data that accurately reflect affluent households (Piketty and Saez, 2006;
CBO, 2009), however, the annual income measures only include the flow realized from wealth
(capital) in any one year.4 In addition, the higher one goes in the income or earnings distribution,
available. In their empirical analysis of the US, Michelangeli et al (2009) find some support for rising concentration,
but not as great as that suggested when using data for household net worth. 3 Only a few recent analysts doubt there has been a widespread increase in inequality that can be generally attributed
to the growth of high incomes (Reynolds, 2007; Tatom, 2007; but see critiqued in Burtless, 2007). 4 Unearned income from transfers, public and private, also accrues but accounts for under 10 percent of incomes.
9
the more likely one is to find high rates of turnover in top incomes from year to year. Indeed,
advocates of high American income mobility point out that the top 1 percent of income earners
have 70 percent turnover rates year-to-year (Cox and Alm, 1999).
This problem is exacerbated by the fact that powerful income recipients can choose the
form and timeframe in which their compensation is paid, e.g., for tax reasons (Auten and Carroll,
1999; Gruber and Saez, 2002). For instance, the two founders of Google, in a widely reported
press story, took $1 each in earnings in 2005. Of course, each one also exercised less highly
taxed stock options, which left them with $1.0 billion or more in ‗asset incomes‘ in that year
(Ackerman, 2006). Whether for reasons of tax and estate planning, or simple accumulation, the
large majority of the gains from wealth, are not realized annually.5
The question we address is in this paper is how to add this income to household
distributional micro-data, and determine to whom did this property or capital income accrue?
The key to pulling these disparate sources and trends in economic well being together is a more
full accounting of annual income from wealth, whether realized or not. Indeed, we believe that
much of what has been interpreted as ―consumption from wealth‖ is not drawing down wealth
stocks at all, but comes from spending out of accretions to wealth (see Love and Smith, 2007, for
older households; and Sierminska and Takhtamanova, 2006, for an international comparison).
Similarly, the declines in US savings rates over many years leading up to the recession were
largely composed of spending from accumulated assets, especially owned homes and other
appreciating assets. While the run-up in home values and dividends received through 2007 fueled
consumer spending (e.g. Baker, et al, 2006), steep declines in housing values since have
diminished consumption due to a decrease in wealth stocks (Glick and Lansing, 2010) and the
5 The sporadic realization of growing incomes from wealth in both the personal and corporate sector, has led to
serious miss-estimates of both individual and corporate income tax revenues at the federal and state level for the past
decade, and especially in recent years (e.g., Schwabish, 2006, CBO, 2006a; 2006b; 2006b; Orszag 2007).
10
savings rate has risen. Clearly, wealth increasingly matters for consumption as well as for
income.
The idea of accounting for income from wealth as well as income from earnings and
other sources is not new (see Weisbrod and Hansen, 1968; Taussig, 1973), and has been used
recently by Wolff and Zacharias (2006a; 2006b) and Haveman, et al., (2006) in some fashion, to
study inequality trends in the 1980s and 1990s.6 Nevertheless, it is clearly time for a reappraisal
given recent seismic changes in overall labor and capital income flows.
II. Income Theory and Methodology
There are many definitions of personal (macro) and household (micro) income from both
a ―sources‖ and ―uses‖ perspective. According to the most popular theoretical measure of
income, the Haig-Simons (H-S) income definition, income (I) is equal to consumption (C) and
the change in net worth ( ∆NW ) realized over the income accounting period. So defined, H-S
income is a measure of potential consumption or the amount one could consume without
changing one‘s total net worth (one‘s stock of assets or debts). Thus according to a ―uses ―of
income definition:
(1) I = C + ∆NW
From the functional or ―sources‖ side of income, we can arrive at the same measure by
adding together income from earnings (E, including self-employment income), income from
capital (KI, including capital gains plus other income from wealth), plus net transfers (NT, which
6 Wolff and Zacharias (2006a, 2006b) and Haveman, et al. (2006; 2007) use an annuity-based measure of inequality
that assumes that all persons, including high income-high wealth persons consume all wealth before they die. Such
measures imply the need for assumptions on discount rates, life expectancy and other variables, and they therefore
assume no bequest or inter-vivos transfer behaviors and they ignore the observed behavior of the rich (e.g. see
Goolsbee, 2007; Carroll, 2000). We prefer a more straightforward estimate of income from wealth using current and
long run rates of return on existing assets. This seems closer to Haig Simons income in terms of capacity to
consume, without the extra baggage entailed with the annuity estimates which necessarily suggest higher incomes
for much older persons, by design
11
includes those received minus those paid, whether private or public in nature), resulting in the
following definition:
(2) I = E + KI + NT
If we ignore NT for now, and divide self-employment income, into income from labor and
capital, we are left with the macroeconomists‘ functional distribution of income.
The key element that is included above but largely missing in most estimates of both
micro and macro estimates of income distribution is the distribution of income from capital.
Despite long-standing interest in labor and capital ―factor shares,‖ macroeconomists (e.g.,
Goldfarb and Leonard, 2005; Guscina, 2006) and microeconomists who study distribution are
both seemingly content with using data where only a small fraction of income from capital is
measured. Interest, rent and dividends received are reported in most micro data based income
definitions such as the one used by the Census Bureau. Capital gains and losses (KG, including
those from realized stock options) and royalties, are counted in other income definitions such as
that used by the CBO (2009) and by Federal Reserve Bank in the SCF income distribution
measure.7
However, the large majority of capital income (KI) accrues to persons but is never
realized (and is therefore not counted in any given year). This includes imputed rental flows for
owner occupied housing; business savings in the form of corporate and non-corporate retained
earnings; and unrealized capital gains. Much of this income stays with the firm that utilizes
capital and is not realized by the owners of these assets (except as it is reflected the value of their
enterprise, either self owned or as shares of corporate stock).
7 Indeed Pryor (2007) attests to the importance of interest rent and dividends in resizing economic inequality using
the PSID.
12
Thus, we define ‗more complete income‘ (or MCI) as follows. We retain earnings and net
transfers (E, NT), and maintain that portion of capital income (KI) received as capital gains and
royalties (KG). But we then subtract reported interest, rent and dividends (IRD) while adding
back in an imputed return to all forms of net worth, or ―imputed capital income‖ (IKI). Thus, we
impute interest rent and dividends to owners of assets and forego the amounts actually reported
by respondents.8 This produces:
(3) MCI=E+ NT+ (KG –IRD + IKI)
Indeed this more complete definition of capital income (KI, below) comes close to measuring the
concept of ‗∆NW‘ that intrigued both Haig and Simons:
(4) KI= KG –IRD + IKI
MCI is an incomplete concept of income as we are unable to measure such items as
employer benefits, pension fund accruals not counted as personal wealth such as defined benefit
pension plans (though pension flows for elders are counted as transfers received), or unrealized
stock options and other promised contractual benefits (‗golden parachutes‘) which are not yet
exercised or received.9
II.a. Developing More Complete Income (MCI) estimates with the SCF
We calculate MCI using the Survey of Consumer Finances, a nationally representative,
triennial survey that includes an over-sample of wealthy households that are underrepresented in
most standard surveys. The SCF contains high quality, detailed information on household assets
8 Reported interest, rent and dividends in the CPS is barely more than half the aggregate amount which other data
suggests ought to be reported (CBO, 2006b). 9 Assets in defined benefit pensions are problematic both because of the potential not to be collected and because of
back loading in benefit determination. We are less worried about the distributional consequences because most such
pensions accrue to the top end of the income distribution and therefore do not affect lower incomes. Our analyses
also ignore non-cash public sector benefits such as those provided by health, education, and the taxes used to pay for
them (see Garfinkel, et al, 2006, on the latter). While these benefits are especially important for low income persons,
they pale in comparison to the levels of imputed income from assets for the large majority of households, especially
middle and high income units. Hence, while MCI helps us better understand the impact and importance of residual
wealth and the way it affects public and private finances and inequality, it does not represent a complete accounting
of all flows of income from all sources.
13
as well as income.10
There are 16 broad asset classes, including stocks, bonds, mutual funds,
home-equity, residential real estate, and business assets, as well six broad classes of debt. The
data include an income definition (SCF income) that is broader than the standard Census money
income definition. SCF income includes wages, self-employment and business income, taxable
and tax-exempt interest, dividends, realized capital gains, food stamps and other support
programs provided by the government, pension income and withdrawals from retirement
accounts, Social Security income, alimony, and other support payments, and miscellaneous
sources of income.11
Income net wealth (―income less capital‖) is calculated by subtracting realized income
from capital from the SCF income definition. Gains from the sale of an asset (capital gains),
however, are retained in the income measure.12
After removing income from capital from SCF
income, flows to assets are imputed for the full range of assets measured in the SCF data. In
calculating the implicit return on various assets, we employ two techniques: first we apply ―short
term‖ (3 year) average rates of return to 22 specific asset/debt types in each of our 8 income
years; and then also ―long run‖ 30 year average returns over the entire period.13
These long run
10
The sample size for the surveys conducted in 2006, 2003 and 2001 was approximately 4,500 households, a slight
increase over that in previous years. 11
Household weights contained in the SCF data are used in all of the calculations. 12
To the extent that capital gains realized in year X are not consumed, but reinvested and emerge as assets in year
X+1, retaining capital gains in the income less capital measure introduces the possibility of ―double counting‖ into
the MCI concept. In any case, this decision to include or exclude realized capital gains has a negligible effect on the
results presented here. And, to the extent that these gains are consumed and not re-invested, excluding them would
understate capital incomes. 13
Other analysts have described the limitations of standard measures of income for welfare and inequality analysis,
and proposed solutions by supplementing income with wealth, as much as a half-century ago. Weisbrod and Hansen
(1968) and Taussig (1973) added the annuity value of net-worth to current income to develop measures they
respectively called ―income-net worth‖ and ―comprehensive income.‖ In more recent work, Wolff and (2006a,
2006b) and Goolsbee, (2007) use the annuity approach for non-housing wealth and impute rental income for
homeowners. There are a number of additional differences between the approach used in this paper and the one used
by Wolff and Zacharias (WZ). WZ use SCF for 1983-2001, we use data for 1989-2007. WZ do not conduct any
after-tax analysis. For their inequality measures, WZ rely primarily on the Gini index and income shares of different
percentile groupings (top 10%, top 1%, etc.) We use Ginis as well, but rely primarily on ratios of key percentiles of
the income distribution (99/50, 95/50, 99/90, etc.) because we find that the biggest impact from using the more
complete income approach is found at the very highest income levels and does not have as great of an impact in the
14
rates allow us to separate more permanent long run returns from more volatile short run changes,
and to assess more smooth trends in income from assets. They also allow us to test the sensitivity
of our results to various assumed rates of return.
Separate rates of return were calculated for stocks, bonds, and housing assets, based,
respectively, on the Dow Jones Industrial Average, 10-year US Treasury notes, and the House
Price Index of the Federal Housing Finance Agency (FHFA). In addition, flows to assets are
calculated gross of the inflation rate (CPI-U), while some flows are based on the average of two
different types of return (the average of the return to stocks and bonds, for example). The actual
rates used to impute these flows are included in Appendix Tables 1 and 2. The complete details
on the construction of MCI, including how taxes are calculated for the various components of
MCI so that we can create pre-tax as well as after-tax inequality measures, are provided in the
Technical Appendix.14
The following additive series of combined capital income flows are added to income, net
of reported interest rent and dividends, in the order specified below:
“plus finance” adds imputed flows to directly held stocks, stock mutual funds,
combination mutual funds, bonds, other bond mutual funds, savings bonds, government
bond mutual funds, and tax free bond mutual funds, as well as ―other managed assets,‖
such as trusts and annuities to “income less capital‖;
“plus retire” adds flows to ―quasi-liquid retirement accounts,‖ such as IRA/Keoghs and
account-type pensions to ―plus finance‖;
“plus home” adds flows to owner-occupied home equity to ―plus retire‖;
“plus oth invest‖ adds flows to investment real estate equity, transaction accounts,
certificates of deposit (CDs) and the cash value of whole life insurance to ―plus home‖;
“plus business” adds flows to other business assets and vehicles―only vehicles worth
more than $50,000―to ―plus oth invest‖;
MCI subtracts flows to non real estate debt, including credit card debt, installment loans,
and other debt from ―plus business‖―after replacing observations, where ―plus
business‖ value incomes were below SCF income with the SCF income value.
Gini. In contrast to prior annuity approaches, WZ assign different rates of return to the different asset types that they
annuitize. These rates are long-run returns covering 1960-2000, and generally based on federal Flow of Funds data. 14
We take no account of the amounts of income, which might have been shifted from a heavily taxed form,
earnings, to another less heavily taxed form, capital gains or dividends, for instance (Lemieux, et. al., 2007).
15
Separate estimates for each of these income concepts are created using both long-run (30-year)
averages and short-run (3-year) time specific rates. The long-run rates are based on the average
annual return between 1977 and 2007, with the same long run rate applied to each year of SCF
data―1989, 1992, 1995, 1998, 2001, 2004, 2007, and projections of the data into 2009.
We also explore an alternative treatment of the vehicle assets, computing a service flow
to vehicle ownership, following Slesnick (1994).15
We consider how modifying treatment of this
asset which is particularly important for middle and low-income households influences levels
and trends in inequality. For SCF income, MCI, and all of its components, we calculate a variety
of standard distributional measures, including the Gini Index, ratios or key income percentiles
(including, for example, the 99/50, 90/50, and the 10/50), in addition to income shares held by
the top 1, 5, and 10 percent of the distribution.
II.b. Projecting SCF into 2009
The next round of the SCF (the eventual SCF 2010) will reflect economic conditions in
2009, but will not available until early 2011. Since the economy entered into a deep recession
after 2007, heavily impacting earnings as well as stock markets and housing values, the portrait
of inequality in the most recently available data cannot be expected to reflect current conditions.
In order to present estimates of inequality that reflect the impacts of the ―Great Recession,‖ we
have projected the data from 2007 SCF data into 2009. These projections are based on income
data from the BEA National Income and Product Accounts, asset data from the Federal Reserve
Board Flow of Funds data, and earnings data from the Current Population Survey.
The income and asset categories used to calculate MCI are adjusted according to the
percent change observed in these same categories between the last two quarters of 2007 and
2009. The changes by income and asset category, and the detailed source of each are displayed in
15
See also ERS (2010).
16
Appendix Table 2A. Changes over this period for the stock market reflect not just the decline in
the total market capitalization that started at the end of 2007, but some of the rebound in market
value since the first quarter of 2009. Changes in annual earnings are allowed to vary by
education and industry class, reflecting – at least in part – how the labor markets of different
demographic groups have been impacted by the Great Recession, as described by Engemann and
Wall (2009).16
The earnings measures in the SCF are adjusted based changes in total weekly
earnings between the first eleven months of 2007 and 2009. The change in earnings is calculated
for twenty separate industry-education cells, and reflects the combined impact of changes in
employment, hours, and wages (Appendix Table 2B).17
Not adjusted for inflation, total earnings
declined for most workers with less than a college degree. Total earnings of workers with a high
school diploma or more education rose between 2007 and 2009, but at a rate less than inflation.
Total earnings increased for workers with a college degree in all six industry groups, but less
than inflation in three of those.
Fewer sets of results are calculated for the 2009 projected incomes. Partly this is a result
of not being able to apply short-run rates to data that are themselves projected using changes in
assets and income categories that are themselves functions of short-run rates of return. But, it is
also the case since some of the tables and figures in the paper are driven by the demographic
composition of the population, which is not modified in the projection to 2009.
III. Results
We begin by tracing how the addition of unrealized capital income changes the
distribution of income, in both tables and figures. Then we look at after-tax income and finally
16
We also know that the distribution of housing wealth is not equally distributed across the population, but exhibits
considerable regional variation (Carson and Dastrup, 2009). Because of the sample size and absence of sub-national
geographic identifiers, we are only able to project an average change in housing wealth across the entire country. 17
Changes in weeks worked between 2007 and 2009, because of temporary layoffs or furloughs, will not be
reflected in our measure of earnings changes.
17
examine levels and trends in various income percentiles and the share of final income that is
either from wealth (capital) or labor. We also briefly explore the demographic profile of high-
MCI households.
III.a. From SCF Income to MCI
We begin with Table 2 and Figure 1 where we apply the long run rates of return to
various asset types and chart the way in which this process changes mean and median income in
2006-2007, as well as the 99th
, 95th
, 90th
and 10th
percentiles (and the Gini inequality measure).
As the figures reviewed in Table 1 suggest, capital income makes a great deal of difference to
correctly measured income in the United States. Subtracting some capital income from SCF
gross income (―less capital‖) reduces the mean and median, but as we successively add wealth-
related income components in Table 2, both measures change dramatically. Moving from SCF
income to MCI, mean income rises by 31 percent and the median by 16 percent. The biggest
changes come from stocks; imputed rent on owned homes; and business assets. Owned homes
(―plus home‖) affects large changes in both mean and median as housing is the quintessential
‗middle class asset‘ and is the only capital income flow which significantly boost the median.
Stocks and bonds (―plus finance‖) and business assets (―plus business‖) have larger affects on
the mean due to the skewed distribution of returns accruing mostly to high MCI units. Indeed,
the 99th
, 95th
and 90th
percentiles rise by 49, 41 and 32 percent respectively in 2007 dollars from
SCF to MCI. In contrast, the 10th
percentile increases only by 17 percent across these same
measures. When we take into account, the changes in the medians, the relative inequality
measures, the 99/50, 95/50 and 90/50 ratios still rise by 28, 21 and 13 percent respectively. The
10/50 ratio is the same in SCF income and MCI. The correction of negatives and the subtraction
of debts, reflected in the difference between ‗plus business‘ and MCI, seem to have little effect
on the overall results.
18
In numerical terms, households at the 10th percentile of MCI have incomes of $14,397
(Table 2) and net assets of $23,112 (Appendix Table 4). Income from wealth increases SCF
income by only $2,057 at the 10th
percentile. This is in contrast with MCI and net worth values
of $185,892 and $864,138 at the 90th
percentile, where capital income is $45,005 in 2007. At the
median MCI level of $55,014, a household has a net worth of $152,491 and a gain of $7,709.
However, at the 99th
percentile of MCI, where MCI is $1,031,528, net worth is over $6.5 million
and SCF incomes in 2007 are increased by $338,000 in moving to MCI. Table 3 does the same
with short run rates of return, with very similar results because short-run returns in 2006-07 are
very close to the long-run rates.
The dramatic nature and extent of these changes are easier seen in Figure 1. The mean
and median values on the right side show steady increases, especially for ―plus home‖ at the
median where the appreciation of owned homes leads to a jump from one plateau to another. In
contrast, the mean income rises steadily with big jumps as noted above and smaller changes at
other definitional points. The 95th
and 90th
percentiles also rise relative to the median. The
increases are most dramatic at the very top of the distribution where the bars show that the 99/50
ratio starts below 15 for SCF income and rises to almost 19 for MCI, with the jump mostly due
to business assets and ―other investments.‖ Hence gains from income from wealth accrue largely
to the very top of the income distribution, even after we re-rank incomes with each successive
component of wealth (or finally, debt), and compare incomes to the median household.18
Table 4 shows the impacts of moving to MCI in the 2009 projected income – using long-
18
The MCI rich are similar to, but not the same as the ‗high income‘ units studied by others. For instance, while 79
percent of the same households are counted in the top one percent for both SCF income and MCI, 84 percent of the
same units are in the top 10 percent. These percents have fallen over the past 18 years as well. In 1989, the overlap
was 83 percent in the top centile and 89 percent in the top decile. Hence the top end of the MCI is increasingly
divergent from the top end of the‘ high income‘ sample. As the value of assets continues to appreciate in the longer
run, and as the fraction of income from capital grows relative to labor, we expect that the top centiles in each
distribution will increasingly diverge.
19
run rates. The SCF incomes are very similar to levels from 2007, slightly lower at the mean and
median and at the 99th
, 95th
and 90th
percentiles, but slightly higher at the 10th
percentile. MCI
incomes, however, are considerably lower for most groups in 2009. Moving from SCF income to
MCI raises the mean and median by 27 and 15 percent, respectively, compared to 31 percent and
16 percent in 2007. Adding in the imputed flows to equity in owner-occupied residential real
estate (―plus home‖) has very little impact on income at either the mean or the median, reflecting
the huge national losses in housing values. Moving to MCI raised the 99th
percentile by 43
percent in 2009, but 49 percent in 2007.
III.b. Including Taxes
The after-tax changes, using 2007 SCF data, are summarized in Table 5. We employ the
NBER TAXSIM model to estimate taxes, given existing, and advantaged, rates for taxable
property income.19
Indeed, while including taxes considerably reduces the incomes of high-
income households (MCI declines about $120,000 for the 99th
percentile after including taxes),
the percentage gains from adding wealth are even greater in after tax terms at the highest income
levels. The 99th
percentile of after tax income rises by 75 percent compared to a 49 percent
change for the before tax incomes (Table 2). These results also confirm that after-tax inequality
is lower than pre-tax inequality, with the 99/50 ratio for MCI (short-term rates) falling from 18.8
to 16.4 after including taxes. The Gini index of MCI falls from .608 to .579 after taxes.20
III.c. Trends in Income Inequality for Key Income Definitions
So far, we have discovered that at any point in time, accounting for income from wealth
drastically increases both the level of income and the inequality of income. To see how the trend
has evolved over the last 20 years, we calculate similar before tax figures for 1988-89, 1991-92,
19
For a discussion of the TAXSIM model see Feenberg and Coutts (1993). 20
We do not calculate the effects of ‗privileged‘ types of taxable income (capital gains, dividends, and housing
sales) on the composition of pre-tax income.
20
1994-95, 1997-98, 2000-01, and 2003-04.21
We prefer the longer run rates when calculating
trends, but figures and tables using short-run rates are available from the authors. Results for
these earlier years show much the same pattern as we saw above in 2006-2007 with few
changes.22
The six graphs in Figure 2 summarize the trend in key income definitions and
component comparisons, using long run rates, over that period. First, MCI is at the top of every
set of lines (except the 10/50 ratio where moving from SCF to MCI has little impact).
While SCF and MCI follow similar patterns at the top of the distribution, the gap between
MCI and SCF income is especially apparent for the 99/50 and 95/50 ratios and for the pattern of
mean incomes. Thus, the trend in inequality is stronger with a more complete (vs. a less
complete) income measure. At the bottom, we see that mean and median incomes both rise over
the period for each income definition, with stagnant periods during previous recessions (early
1990s and 2001), but outright declines in the Great Recession. The 90-50 ratios show little trend,
suggesting most gains over the period are concentrated at the top of the distribution. The dips in
the 99/50 ratio in 2003 and again in 2009 reflect the collapse of the stock (and housing in 2009)
market in those periods.
Adding imputed flows for financial wealth (―plus finance‖) to income ―less capital‖
leaves the 99/50 ratio very similar to SCF income. Adding housing wealth (―plus home‖)
produces little change in the 99/50 and 95/50 ratios, but accounts for the bulk of the change at
the median (Panel F) and a large portion of the change in the mean (Panel E). The bulk of the
gap between SCF income and MCI in the 99/50 ratio is a result of one of the final elements of
21
Figures illustrating equivalent trends for after-tax income were included in previous draft, and are available from
the authors. 22
Data for years before 2007 are contained both in an earlier version of the paper, available at:
http://www.irp.wisc.edu/aboutirp/people/affiliates/Smeeding/14-INCOME-FROM-WEALTH_6_21_07.pdf , as well
as an additional series of data tables: http://www.irp.wisc.edu/aboutirp/people/affiliates/Smeeding/14b-Appendix-
tables-available-from-authors-Jun-7-2007.pdf.
21
MCI, imputed flows to business wealth. The relevance of business wealth shows up in the means
(Panel E) and the 99/50 ratio (Panel A), but not the other trend statistics.23
In general, the trends presented in Figure 2 suggest the effects of adding income from
wealth follow a similar pattern of rising inequality as seen in the SCF income as well as other
measures of income inequality over this period (e.g. Smeeding, 2005; CBO 2010). While
inequality is higher in any given year for MCI income than SCF income, the 95/50 and 90/50
ratios follow the same upward trend as the SCF income (Panels B, C). For the very top of the
distribution, however, the inclusion of income from wealth results in a more dramatic rise in
inequality (Panel A). The 99/50 ratio rises 57 percent between 1988-89 and 2006-07 in the SCF
income measure, but it increases 64 percent for MCI. Therefore, while Wolff and Zacharias
(2006a; 2006b) show that an augmented measure of wealth results in about the same rise in
inequality as traditional measures of money income, our approach suggests that inequality
appears to rise even higher when we use more complete measures of income.
Projections to 2009 suggest that the run-up in inequality between 1989 and 2007 was
halted by the Great Recession. The 99/50 ratio declined from 18.8 in 2007 to 17.5 in 2009 using
MCI and by a smaller amount using SCF income. Other MCI-based measures of inequality
(Appendix Table 3) also declined over this period; the Gini index dropped from .608 to .600.
Most of these measures, however, also show that inequality remains at high levels.
III.d. Percentile Growth in Incomes
Figure 3 (Panels A and B) summarizes the 1989 to 2007 growth rates for SCF income
and MCI across the entire distribution. The growth in MCI is greater than SCF income for all
households above the 40th
percentile of the income distribution. Over most of the income
distribution, the importance of moving to MCI appears to be roughly constant with the gap in
23
Equivalent figures using short run rates show essentially the same patterns.
22
growth rates fluctuating between 10 and 20 percentage points (Panel B). At the top of the
distribution, however, the gap in growth rates increases dramatically. For the top three percent of
the income distribution growth in MCI is more than 30 percent higher than SCF. For the 99th
percentile MCI growth was 35 percentage points faster than SCF income. Hence, the inclusion of
income from wealth results in a rising inequality trend, when the measure of inequality contrasts
the highest-income households with any other grouping.
III.e. Tends in the Income Share of Top-Income Households
There are several sets of estimated income trends amongst the rich to which we can
compare our results. In Figure 4, we compare MCI shares of total income using long run rates to
those found in three other studies: the Wolff-Zacharias (WZ, 2006a; 2006b) annuity value
measures of income net worth; the CBO (2010) income after taxes and benefits including capital
gains series; and those compiled in the ‗top income‘ papers of Piketty and Saez (PS, 2003; 2007).
We have plotted the shares, and have calculated the trends and the slopes of each line.
First we note that the top one percent shares using MCI are roughly in line with those of
PS and WZ (Panel A). And, while Reynolds (2007) and Tatom (2007) have criticized the PS
numbers because more of high income is not reported for tax reasons, our MCI measure avoids
this problem, as we include unrealized and therefore untaxed income from wealth, and our shares
are at least as high if not higher. For the top one percent, all lines rise over the period, suggesting
an increase in share for either the 1989-2007 or 1989—2001 periods. For the 1989-2001 period
MCI and PS had the steepest slopes for growth in the income share of the top one percent, well
above the rates of increase in either the CBO or WZ figures. Between 1989 and 2007, though,
the MCI slope was close to, but slightly smaller than, the CBO slope, and considerably smaller
than the PS slope. To varying degrees, all of the series show a rising share at the very top.
23
For the next tier of top-income households – the 95th
to 99th
percentiles of the distribution
(Panel B) – the PS and CBO series show relatively low growth over the period (1989-2001 or
1989-2007). Using MCI, however, the income share of this group rises considerably, as does the
WZ series for 1989-2001. All of the series show slightly declining income shares for the next
tier of top-income households – the 90th
to 95th
percentiles of the distribution (Panel C).
Based on the projections for 2009, MCI shows that the top income shares have declined
slightly since 2007; the income share of the top one percent fell from 22.3 to 21.9 percent
between 2007 and 2009, while the share of the next-highest four percent declined from 18.5 to
18.0 percent. The most recent IRS tax data analyzed by Piketty and Saez also show a decline in
the income share of the top 1 percent between 2007 and 2008, but record a slight uptick in the
share of the next four percent. The crisis in the financial sector and the decline in business assets
appears to have made a small dent in the income share of the highest-income households, but the
shares remain higher than every year before 2007.
III.f. Impact of Vehicle Service Flow
The levels and trends in MCI shown in tables 2, 3, and 4 and figures 1, 2, 3, and 4 do not
impute any flow of income to vehicle assets below $50,000 in value. For low and middle-income
households, however, vehicles are an important asset, and are typically valued well below
$50,000. In 2007, median vehicle value was $13,000. Nearly 90 percent of households had
vehicle assets, and the value at the 90th
percentile was under $41,000. Changing the treatment of
vehicles in MCI, calculating an annual service flow to vehicle ownership following the approach
of Slesnick (1994), suggests somewhat higher levels of MCI, particularly for low-income
households.24
Including vehicle service flow, MCI for at the 99th
percentile was $943,740 in
24
The SCF includes a set of detailed questions about household vehicles, but in the public-use version of the data
only includes the value of the vehicle. SCF staff calculate vehicle value based on the age, make, model, and
condition of the vehicle. To calculate the vehicle service flow, we multiply the vehicle value by the rate of
24
2009, one percent higher than the baseline approach in Table 4. MCI at the 10th
percentile,
though, was $15,538 with vehicle service flow, 5.2 percent higher than baseline MCI. Including
vehicle service flows also modestly changes measures in the level of inequality, but has no
appreciable impact on inequality trends. The Gini index of MCI in 2007, for example, falls from
.600 to .593 after including vehicle service flow. The trend in the Gini index, though, is not
affected (Figure 5, panel A). Adding vehicle service flow to MCI results in similar small changes
in other inequality measures, including the income share of top five percent on households and
the 99/50 income ratio, but leaves the overall trends unchanged (Figure 5, panels B, C).
III.g. Labor vs. Capital Income
A more complete accounting for income from wealth as well as from labor produces
large changes in the functional income distribution. At the top of Figure 6 in Panel A, we see the
SCF traditional micro-data based pattern of household income components. Earned income is
63-70 percent of gross incomes over the period we study. Indeed most authors (e.g., Cowen,
2007; Tatom, 2007) assume that labor income is always about 65-70 percent of total income.
Conventional reported income from interest, rent, dividends and sometimes capital gains is
between 10 and 15 percent of SCF income. ―Other‖ (largely public transfer) net income is 9-15
percent of gross income, while income from capital and Self Employment Business Income
(SEBI) are both no more than 10 percent. This is the standard picture with almost all household
income micro datasets, but the pattern is considerably different when we consider the MCI
distribution (Panel B).
Now, because we assess all capital income in MCI, capital income is both higher than in
panel A, and is also growing from 1988-89 to 2006-07. The capital share of income in MCI rises
depreciation plus the rate of depreciation. Also following Slesnick (1994, 2001), we assume a rate of depreciation or
10 percent.
25
from 30 to 40 percent over this period, with a recession induced dip in 1991-92 and plateau in
2003-04, before falling back to 37 percent in the Great Recession. Over the same period, the
labor share of income falls to 52 percent in 2006-07 percent, before bouncing back to 55 percent
in 2009. ―Other‖ (net transfer) income changes very little.25
Using short-run rates (not shown)
results in very noisy results that fail to show any trend between 1989 and 2007.
These trends, especially using long-run rates, suggest the role of income from wealth is
growing stronger in the US, while labor income is falling in importance. Simply put income from
wealth rises and income from labor falls once we take a more complete view of Haig-Simons
income.
III.h. Labor Shares at the Top of the Distribution
Similar to Wolff and Zacharias (2006a; 2006b) we find that our expanded measure of
income using the SCF fails to support Piketty and Saez‘s (2003, 2006) finding of the rising
importance of income from labor. Using federal tax return data, Piketty and Saez document a
rising labor share of total money income for high-income households. Using the expanded
income definition of MCI, we find that income from wealth represents the largest share of MCI
at the top of the distribution and that the wealth share is rising.26
Figure 7 shows the share composition of MCI over the entire distribution. For the lowest
MCI households labor and capital combined represent less than one third of total MCI in 2007,
but for the highest MCI households capital income alone constitutes more than half of MCI
(Panel A). The trend comparisons (Panel B) suggest that capital income represents the largest
portion of MCI for the top few percentiles, and the capital share increased between 1989 and
25
The estimates of labor share exclude the non-pension portion of total non-wage compensation. Adding in
employer subsides for health care , the one large and ignored element of compensation might reduce the trend
slightly, but it would not change the qualitative conclusion that the long term capital share is rising 26
Lemieux, et al. (2007) describe how performance-based or incentive based pay has increasingly driven the income
share of the top centime, but these same annual performance pay increases are no doubt also driving accumulated
wealth at the tip of the MCI distribution in recent years, but with a one year or longer lag.
26
2007 for the top five percentiles. For the top one percent of the MCI distribution, the capital
share rose from 39 percent of MCI in 1989 to 53 percent by 2007.
The labor share of MCI, conversely, has declined at the top of the distribution. Figure 8
shows the labor share of income for top-income households using both SCF Income (Panel A)
and MCI (Panel B.) Using SCF income, the labor share of income the top one percent has risen,
though not steadily, since 1989. Using MCI, the labor share declined between 1989 and 2007 for
the top one percent as well as the next four percent, before rising in the ―Great Recession.‖
III.i. Who are the Rich?
The demographic profile of households by MCI class (Table 6) shows that, relative to
other households, high MCI households are older, better educated, more likely to be white and
married, more likely to be self-employed or in a partnership, and are disproportionately grouped
in managerial and professional occupations. Nearly 92 percent of households in the top one
percent were headed by non-Hispanic whites and 92 percent were married, compared to nearly
69 percent and 55 percent, respectively, for the bottom 90 percent of households.27
Age alone is
not a terribly good predictor of high wealth as nearly 42 percent of the group in the top 1 percent
have children under age 18, little different from the bottom 90 percent of households.
The educational and occupational differences between high MCI households and the
general population are quite striking. Nearly 9 of 10 (87 percent) household heads in the top 1
percent of MCI had at least a college degree compared to 30 percent among the bottom 90
percent. Nearly half (47 percent) of working households in the top 1 percent of MCI had at least
some post-graduate education.28
Hence accumulation of human capital is indirectly linked to
income from wealth. More than 88 percent of household heads in the top 1 percent of the MCI
27
These relationships include legally married couples and other couples that are ―partners.‖ 28
Results in expanded tables available from the authors.
27
distribution were in the managerial and professional occupation class, and 45 percent were self-
employed or a partner in a firm, compared to just 34 percent and 8 percent respectively for the
bottom 90 percent of the distribution. Moreover, nearly half (46 percent) of working households
in the top 1 percent were self-employed/partner in a managerial and professional occupation.
High MCI households are a varied lot in certain respects, but their education-occupation-
industry profile suggests a large concentration of managers, business owners and entrepreneurs.
Human capital is important to high MCI, but it appears most successful when combined with
employment or investment in partnerships, self held companies, and high-level management
responsibilities. High MCI households are not especially aged and almost half of high MCI
families still have children under age 18. A more thorough treatment of demographics, including
means, medians, and distributional breakdowns by age, family composition, and ethnicity for
SCF income and MCI, as well as a consideration of the influence of population ageing on
inequality is included in previous drafts.29
IV. Summary/Discussion
Augmenting the standard income definition with flows imputed to assets results in greater
resources for households across the distribution. MCI exceeds SCF income by 16 percent at the
median and 17 percent at the 10th
percentile of the distribution (for 2007). Accounting for the
flow of services provided by, and opportunity to consume represented by, assets suggests
households are better off than if we only consider more narrow definitions of income, but these
resources are also exposed to fluctuations in the stock and housing markets as well as the labor
market. Median income declined $1,000 between 2007 and 2009 using SCF income, but $1,650
using MCI.
29
Available at: http://www.irp.wisc.edu/aboutirp/people/affiliates/Smeeding/14-INCOME-FROM-
WEALTH_6_21_07.pdf.
28
Broadening the income concept to incorporate the benefits of wealth and asset ownership
also results in even larger gains at the top of the distribution; at the 99th
percentile MCI was 49
percent larger than SCF income. At any point in time, income inequality measures are larger
using MCI than standard income concepts – the 95/50 ratio in 2007 was 5.3 using MCI
compared to 4.4 for SCF income. And the rise in the concentration of income among the top few
deciles of the distribution in recent decades appears even greater using this broader income
concept – between 1989 and 2007 the 99/50 ratio rose 63 percent (from 11.5 to 17.5) for MCI,
but only 57 percent using SCF income.
We have documented these increases in inequality using Gini indexes, ratios of key
percentiles (95/50, etc.), and income shares of top-income groups. There are, however, other
inequality metrics that have been developed in recent years specifically to address distributional
questions at the top of the income distribution, including the ―affluence measures‖ proposed by
Peichl, et al (2010) and the approach to ranking intersecting Lorenz curves developed by
Aaberge (2009). These measures will be explored in future work with the MCI concept.
The income trends we are documenting are not simply across the distribution increasingly
toward households with very high incomes, but also across sources of income, from labor to
capital income. High-income families are not increasingly represented by high earners, as Piketty
and Saez (2003; 2006) argue; instead high earners and other asset-rich households are building
up assets and accumulating high-unmeasured incomes from these assets. MCI brings out these
patterns in some detail.
Not unlike the Medici period in Italy, this ―Richistan‖ (Frank, 2007) pattern is definitely
at work in the early 21st century where flat earnings below the 80
th percentile and falling median
incomes for the non-elderly have drawn repeated questions about where the nation‘s productivity
gains have gone. (Gordon and Dew-Becker, 2005; Mishel et al, 2005; Lemieux, et al., 2007;
29
Aron-Dine and Shapiro, 2006). The answer is that they went to, and remain in, higher value
assets, including higher value corporate assets, proprietor‘s incomes, net interest and profits
(which drive up stock and bond market returns and the value of business equity).
And, the US is not alone in this situation, as OECD figures reported by Porter (2006),
Glynn, (2009) and Guscina (2006) suggest that the labor share of total income has fallen in most
rich OECD nations over the 1990-2004 period. Indeed the labor share in Germany and Japan fell
by even more than in the United States over this period, while at the same time; the German
trend has been increasingly for market incomes to accrue to the highest income households
(Bach, Corneo and Steiner, 2007). In addition, concentration of wealth is on the rise in Europe as
well as in the United States (Atkinson, 2006).
Institutional and economic change has created a greater emphasis on worldwide ‗free
market‘ capitalism, high returns to the entrepreneurs—the inventors and creative users of capital
(Acemogolou, 2002). These changes have been combined with tax advantages for both capital
income and high incomes, and have led to the worsening of the social and political position of
labor more generally (Levy and Temin, 2007). All of these factors have contributed to the shift to
higher capital vs. labor income. Ever greater global trade and further technological change
should only intensify these changes (Blinder, 2007; Freeman, 2007). While some claim labor
incomes will rise more in the future than will capital incomes due to world population aging
(Krueger and Ludwig, 2006), others see high and rising returns to asset holdings for those with
productive assets such as pension savings (Poterba, et al., 2007a; 2007b; Love and Smith, 2007).
Indeed while human capital and technology are ―racing‖ for higher income shares (Goldin and
Katz, 2006), technology and the entrepreneurs who own and deploy such capital are currently
winning the race, and are increasingly likely to receive higher rewards in a world of mobile
capital and workers (see also Freeman, 2007).
30
Technical Appendix: Constructing MCI and Adding Taxes
Income net wealth (―income less capital‖) is calculated by subtracting realized income
from capital from the SCF income definition. 30
Hence, reported interest, rents and dividends are
excluded in the given income year. Further, capital gains and royalties are also excluded in
counting income ―less capital‖ to avoid double counting, as we will be imputing returns to these
assets to the extent that these 2006 gains and royalties have been invested in other assets by
2007. To the extent that these gains and royalties are consumed and not re-invested, we will
underestimate capital incomes in this process.
In allocating the functional share of income between labor and capital, and further in
accounting for capital income flows, we partition self employment income as follows: in the
cases where self-employment and business income (SEBI) exceeds income from wages, thirty
percent of SEBI is considered a return to capital and is also subtracted from SCF income to
complete ―less capital.‖ In cases where SEBI is less than income from wages, we treat all SEBI
as income from labor. This practice is the same as that employed by others who also split SEBI
into labor and capital components (e.g., see Canberra Report, 2001).
After removing income from capital from SCF income, flows to assets are imputed for
the full range of assets measured in the SCF data. Separate rates of return were applied for
stocks, bonds, and housing assets. Specific rates applied to the assets are based on historic
returns data described in greater detail below. The return to stocks is based on the Dow Jones
Industrial Average. The return to bonds is based on 10-year US Treasury notes. The return to
residential real estate is based on Office of Federal Housing Enterprise Oversight (OFHEO)
House Price Index. In addition, flows to assets are calculated gross of the inflation rate (CPI-U),
while some flows are based on the average of two different types of return (the average of the
return to stocks and bonds, for example). The details are contained in Appendix Table A-1.
The following additive series of combined capital income flows are added to income, net of
reported interest rent and dividends, in the order specified below:
30
Three different versions of the SCF data for each year are used. The household income variable and many of the
broader asset and income definitions as well as key demographic details are available in the ―Extract of the Full
Public Data Set‖ (in Stata) version of the SCF. This version of the data contains the variables used in Federal
Reserve Bulletin article. Detailed asset classes not included in the extract file were accessed through the ―Full Public
Data Set‖ (in Stata). Key variables from the full data set were merged into the extract file. Finally, the full public
access version of the data was accessed a second time in SAS. This was done because the SCF tax programs are
coded in SAS. Use of the SCF tax programs and NBER‘s TAXSIM is discussed in more detail below. (All of these
versions are available at the SCF web site: www.federalreserve.gov/Pubs/oss/oss2/scfindex.html.)
31
“plus finance” adds imputed flows to directly held stocks, stock mutual funds,
combination mutual funds, bonds, other bond mutual funds, savings bonds, government
bond mutual funds, and tax free bond mutual funds, as well as ―other managed assets,‖
such as trusts and annuities to “income less capital‖;
“plus retire” adds flows to ―quasi-liquid retirement accounts,‖ such as IRA/Keoghs and
account-type pensions to ―plus finance‖;
“plus home” adds flows to owner-occupied home equity to ―plus retire‖;
“plus oth invest‖ adds flows to investment real estate equity, transaction accounts,
certificates of deposit (CDs) and the cash value of whole life insurance to ―plus home‖;
“plus business” adds flows to other business assets and vehicles―only vehicles worth
more than $50,000―to ―plus oth invest‖;
MCI subtracts flows to non real estate debt, including credit card debt, installment loans,
and other debt from ―plus business‖―after replacing observations, where ―plus
business‖ value incomes were below SCF income with the SCF income value.31
Separate estimates for each of these income concepts are created using both long-run (30-
year) averages and short-run (3-year) time specific rates. The long-run rates are based on the
average annual return between 1977 and 2007, with the same long run rate applied to each year
of SCF data―1989, 1992, 1995, 1998, 2001, 2004, 2007, and projections of the data into 2009.32
Short-run returns are averages of the three years leading up to the survey year. The short-run
return for income year 1989, for example, is based on the annual average return between 1987
and 1989. Income is from the completed calendar year prior to the survey. Assets are valued at
the time of the survey, completed in the second half of the year. Imputed flows for 1989, for
example, are based on wealth stocks reported between June and December of 199033
.
The long run nominal rates of return for stocks, bonds, housing and inflation are 7, 5, 6
and 3 percent, respectively and are smaller than the 1977-2007 and 1989-2007 averages for this
period. We believe that the long run rates are modest and we know that they reflect estimates
used by others. For instance, the 4 percent real return for stocks (7 percent minus 3 percent
inflation adjustment) is the same as that used by the Social Security Advisory Board to score the
net effects of investing Social Security funds in the private equities market. Also the six percent
31
This adjustment was made on account of households with negative incomes even after imputation of flows to all
assets. These households had large trust and royalty income, but experienced negative capital gains income that left
them with relatively low (or zero) SCF income. When the trust and royalty income was subtracted from SCF
income, the result was deeply negative income that dwarfed the imputed flow to their assets. This occurred in less
than three percent of households in the 2003 data. The adjustment has little or no effect on the overall results. 32
The actual long-run rates applied reduced the return to bonds and stocks by roughly 3.0 percentage points to
adjust for annual rates of inflation. See Appendix Table 1 for details. 33
Appendix Tables 1 and 2 include details for the long run and short-run rates of return applied to each income
concept between 1989 and 2007. The year to year short run rates vary by period and asset type (see Appendix Table
2).
32
rate of return for housing is identical to the real return that Case (2010) suggests homeowners
should anticipate after maintenance. Finally, we assume a long run non-housing debt rate of 9
percent. Housing debt is factored in when determining net imputed rent on owned and other
housing equity.
Incorporating Taxes
In addition to the MCI concepts described above, three additional after-tax income
concepts are calculated for each year up through 2007. Taxes for all three are federal income
taxes calculated using the National Bureau of Economic Research (NBER) TAXSIM program.
All of the required input for TAXSIM is generated based on programs developed by Fed
economist Kevin Moore, and is available on the NBER web site.34
The first after-tax concept is simply reported SCF income less taxes, a version of
disposable personal income (dpi). The second concept is income net wealth and net taxes.
Income net wealth is defined as described above (―less capital‖) and the related taxes are
calculated with TAXSIM by eliminating dividend and ―other property‖ income, including
interest, from the input file.35
The final after-tax concept is based on MCI. In this case, the sum
of the imputed flow to assets included in MCI is categorized as dividend income and the taxes
calculated by TAXSIM.36
The resulting federal taxes are subtracted from MCI to create ―MCI
less tax.‖
34
The TAXSIM is available online at: http://www.nber.org/~taxsim/. 35
These are fields 9 and 10 of the TAXSIM input file. 36
In addition these results are also calculated with the imputed flows in MCI classified as ―other property income‖
in TAXSIM. The impacts of this difference are minimal, and only present for 2004 and after.
33
References
Aaberge, Rolf. 2009. ―Ranking Intersecting Lorenz Curves.‖ Social Choice and Welfare 33 (2):
235-259.
Acemoglu, Daron. 2002. ―Technical Change, Inequality, and the Labor Market.‖ Journal of
Economic Literature 40 (1): 7-72.
Ackerman, Elise. 2006. ―Google ‗Business Founder‘ No.1 - 289 Million, Plus Stock Option
Gains.‖ Mercury News.
Aron-Dine, Aviva and Isaac Shapiro. 2006. ―Share of National Income Going to Wages and
Salaries at Record Low in 2006: Share of Income Going to Corporate Profits at Record
High.‖ Washington, DC: Center on Budget and Priorities. Revised March 29, 2007.
http://www.cbpp.org/8-31-06inc.pdf.
Atkinson, Anthony B, Thomas Piketty, and Emmanuel Saez. 2009. ―Top Incomes in the Long
Run of History.‖ NBER Working Paper 15408, October.
Atkinson, Anthony B. 2006. ―Concentration among the Rich.‖ Research Paper No. 2006/151.
Helsinki, Finland: United Nations University-World Institute for Development
Economics Research. December.
Atkinson, Anthony B. 2009. ―Income inequality in historical and comparative perspective: A
graphical overview.‖ Conference on Inequality in a time of Contraction at Stanford
University, November 2009.
Auten, Gerald and Robert Carroll. 1999. ―The Effect of Income Taxes on Household Income.‖
Review of Economics and Statistics 81 (4) (November): 681-693.
Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. 2006. ―The Polarization of the U.S.
Labor Market.‖ American Economic Review 96 (2): 189-194.
Autor, David, Lawrence F. Katz, and Melissa S. Kearney, 2008. ―Trends in U.S. Wage
Inequality: Revising the Revisionists,‖ Review of Economics and Statistics, 90 (2): 300-
23.
Bach, Stefan, Giacomo Corneo, and Viktor Steiner. 2007. ―From Bottom to Top: The Entire
Distribution of Market Income in Germany, 1992-2001.‖ IZA Discussion Papers, 2723.
Institute for the Study of Labor. April.
Baker, Malcolm, Stefan Nagel, and Jeffrey Wurgler. 2006. ―The Effect of Dividends on
Consumption.‖ NBER Working Paper No. 12288. Cambridge, MA: National Bureau of
Economic Research. June.
Blinder, Alan S. 2007. ―How Many US Jobs Might be Offshorable?‖ CEPS Working Paper: No.
142. Princeton, NJ: Center for Economic Policy Studies. March.
34
Brandolini, Andrea and Timothy Smeeding, 2009. ―Income Inequality in Richer and OECD
Countries,‖ in Oxford Handbook of Economic Inequality, W. Salverda, B. Nolan, and T.
M. Smeeding (eds.), 71-100. Oxford, UK: Oxford University Press.
Bureau of Economic Analysis. 2010. National Income and Product Accounts, Table 1.12,
Burkhauser, Richard V., Shuaizhang Feng, Stephen Jenkins, and Jeff Larrimore. 2009. ―Recent
Trends in Top Income Shares in the USA: Reconciling Estimates from March CPS and
IRS Tax Return Data.‖ NBER Working Paper No. 15320. Camrbidge, MA: National
Bureau of Economic Research.
Burtless, Gary. 2007. ―Comments on ‗Has US Income Inequality Really Increased‘.‖ Presented
at Cato Institute.
Canberra Group. 2001. Expert Group on Household Income Statistics: Final Report and
Recommendations. Ottawa, Canada: Statistics Canada.
Carroll, Christopher D. 2000. ―Why Do the Rich Save So Much?‖ In Does Atlas Shrug? The
Economic Consequences of Taxing the Rich, edited by Joel B. Slemrod. Cambridge, MA:
Harvard University Press, 465-484.
Carson, Richard and Samuel Dastrup. 2009. ―After the Fall: An Ex Post Characterization of
Housing Price Declines Across Metropolitan Areas.‖ U.C. San Diego Working Paper,
November 23.
Case, Karl. 2010. ―A Dream House After All.‖ The New York Times. September 1, 2010.
Congressional Budget Office. 2006a. ―How CBO Forecasts Income.‖ Congressional Budget
Office Background Paper. Washington, DC: CBO. August.
Congressional Budget Office. 2006b. ―The Treatment of Federal Receipts and Expenditures in
the National Income and Product Accounts.‖ Washington, DC: CBO. September.
Congressional Budget Office. 2010. ―Historical Effective Federal Tax Rates: 1979 to 2007.‖
Washington, DC: CBO. June.
Cowen, Tyler. 2007. ―Why Is Income Inequality in America So Pronounced? Consider
Education.‖ New York Times, May 17.
Cox, W. Michael and Richard Alm. 1999. Myths of Rich &Poor: Why We’re Better Off Than We
Think. 1st edition. New York: Basic Books.
Davis, Bob and Robert Frank. 2009. ―Income Gap Shrinks at the Expense of the Wealthy.‖ Wall
Street Journal, September 10.
Economic Research Service, 2010. ―Farm Household Well-Being: Comparing Consumption- and
Income-Based Measures,‖ US Department of Agriculture Economic Research Report No.
(ERR-91), February.
35
Engemann, Kristie and Howard Wall. 2009. ―The Effects of Recessions Across Demographic
Groups.‖ Federal Reserve Bank of St. Louis, Working Paper 2009-052A, October.
Feenberg, Daniel and Elisabeth Coutts. 1993. ―An Introduction to the TAXSIM Model.‖ Journal
of Policy Analysis and Management 12 (1): 189-194.
Frank, Robert. 2007. Richistan: A Journey through the American Wealth Boom and the Lives of
the New Rich. NY: Crown Publishers.
Freeman, Richard B. 2007. America Works: Critical Thoughts on the Exceptional U.S. Labor
Market. New York: Russell Sage Foundation.
Garfinkel, Irwin, Lee Rainwater, and Timothy M. Smeeding. 2006. ―A Reexamination of
Welfare State and Inequality in Rich Nations: How In-Kind Transfers and Indirect Taxes
Change the Story.‖ Journal of Policy Analysis and Management 25 (4): 897-919.
Glick, Reuven and Kevin Lansing. 2010. ―Global Household Leverage, House Prices, and
Consumption, Economic Letter #2010-01, Federal Reserve Bank of San Francisco,
January.
Glyn, Andrew, 2009. ―Functional Distribution of Income,‖ in Oxford Handbook of Economic
Inequality, W. Salverda, B. Nolan, and T. M. Smeeding (eds.), 101-125. Oxford, UK:
Oxford University Press.
Goldfarb, Robert S. and Thomas C. Leonard. 2005. ―Inequality of What among Whom? Rival
Conceptions of Distribution in the 20th Century.‖ In Research in the History of Economic
Thought and Methodology, Volume 23-A, edited by Warren J. Samuels, Jeff E. Biddle,
and Ross B. Emmett. Oxford, UK: Elsevier, 75-118.
Goldin, Claudia and Lawrence F. Katz. 2007. ―The Race between Education and Technology:
The Evolution of U.S. Educational Wage Differentials, 1890 to 2005.‖ NBER Working
Paper No. 12984. Cambridge, MA: National Bureau of Economic Research. March.
Goolsbee, Austan. 2007. ―Why Do the Richest People Rarely Intend to Give It All Away?‖ New
York Times, March 1.
Gordon, Robert J. and Ian Dew-Becker. 2005. ―Where did the Productivity Growth Go? Inflation
Dynamics and the Distribution of Income.‖ Brookings Papers on Economic Activity (2):
67-127.
Gruber, Jon and Emmanuel Saez. 2002. ―The Elasticity of Taxable Income: Evidence and
Implications.‖ Journal of Public Economics 84 (1) (April): 1-32.
Guscina, Anastasia. 2006. ―Effects of Globalization on Labor‘s Share in National Income.‖ IMF
Working Paper 06/294. Washington, DC: International Monetary Fund. December.
Haveman, Robert, Karen Holden, Barbara Wolfe, and Shane Sherlund. 2006. ―Do Newly Retired
Workers in the United States Have Sufficient Resources to Maintain Well-Being?‖
Economic Inquiry 44 (2) (April): 249-264.
36
Haveman, Robert, Karen Holden, Barbara Wolfe, and Andrei Romanov. 2007. ―Assessing the
Maintenance of Savings Sufficiency Over the First Decade of Retirement.‖ International
Tax and Public Finance. August.
Heathcote, Jonathan, Fabrizio Perri, and Giovanni L. Violante. 2010a. ―Unequal We Stand: An
Empirical Analysis if Economic Inequality in the US, 1967–2006.‖ Review of Economic
Dynamics 13(1): 15–51.
Heathcote, Jonathan, Fabrizio Perri, and Giovanni L. Violante. 2010b. ―Inequality in Times of
Crisis: Lessons from the Past and a First Look at the Current Recession.‖ VoxEU.org,
February.
Kaplan, Steven N. and Joshua D. Rauh. 2010. ―Wall Street and Main Street: What Contributes to
the Rise in the Highest Incomes?,‖ The Review of Financial Studies, Vol. 23 (3), 1004-
1050.
Katz, Lawrence. 2006. ―Narrowing, Widening, Polarizing: The Evolution of the U.S. Wage
Structure.‖ Presented at Society of Labor Economists Eleventh Annual Meeting,
Cambridge, MA, May.
Krueger, Dirk, Fabrizio Perri, Luigi Pistaferri, and Giovanni L. Violante. 2010. ―Cross-Sectional
Facts for Macroeconomists.‖ Review of Economic Dynamics 13(1): 1–14.
Krueger, Dirk and Alexander Ludwig. 2006. ―On the Consequences of Demographic Change for
Rates of Returns to Capital, and the Distribution of Wealth and Welfare.‖ NBER
Working Paper No. 12453. Cambridge, MA: National Bureau of Economic Research.
August.
Lemieux, Thomas. 2006. ―Increasing Residual Wage Inequality: Composition Effects, Noisy
Data, or Rising Demand for Skill?‖ American Economic Review 96 (3) (June): 461-498.
Lemieux, Thomas, W. Bentley MacLeod, and Daniel Parent. 2007. ―Performance Pay and Wage
Inequality.‖ NBER Working Paper no. 13128. Cambridge, MA: National Bureau of
Economic Research. May.
Leonhardt, David and Geraldine Fabrikant. 2009. ―Rise of the Super-Rich Hits a Sobering Wall.‖
New York Times, August 21.
Levy, Frank and Peter Temin. 2007. ―Inequality and Institutions in 20th Century America.‖
NBER Working Paper No. 13106. Cambridge, MA: National Bureau of Economic
Research. May.
Love, David and Paul Smith. 2007. ―Measuring Dissaving out of Retirement Wealth.‖ (March
2007). Available at SSRN: http://ssrn.com/abstract=968431.
Meyer, Bruce and James Sullivan. 2010. ―Consumption and Income Inequality in the US: 1960-
2008,‖ Working Paper, December 27, 2009.
37
Meyer, Bruce and James Sullivan. 2003. ―Measuring the Well-Being of the Poor Using Income
and Consumption,‖ Journal of Human Resources, 38(S): 1180-1220.
Michelangeli, Alessandra, Eugenio Peluso, and Alain Trannoy. 2009. ―American Baby-Losers?
Robust Indirect Comparison of Affluence Across Generations.‖ ECINEQ Working Paper
2009-133, September.
Mishel, Lawrence, Jared Bernstein, and Sylvia Allegretto. 2005. The State of Working America,
2004/2005. Ithaca, NY: Cornell University Press.
Monea, Emily and Isabel Sawhill. 2009. Simulating the Effect of the “Great Recession” on
Poverty. Washington, DC: Brookings Institution, Center on Children and Families.
September.
NBER. 2010. Business Cycle Expansions and Contractions. Accessed on October 10, 2010, at
http://www.nber.org/cycles.html.
Orszag, Peter R. 2007. ―Letter to the Honorable Kent Conrad on Federal Tax Revenues from
2003 to 2006.‖ Washington, DC: Congressional Budget Office. May.
Parker, Jonathan A., and Annette Vissing-Jorgensen. 2009. ―Who Bears Aggregate Fluctuations
and How?‖ Papers and Proceedings of AEA. May.
Peichl, Andreas, Thilo Schaefer, and Christoph Scheicher. 2010. ―Measuring Richness and
Poverty: A Micro Data Application to Europe and Germany,‖ Review of Income and
Wealth 56 (3) (September): 597-619.
Petev, Ivaylo, Luigi Pistaferri, and Victoria Saporta. 2010. ―Consumption and the Great
Recession.‖ Presented at Stanford Poverty Conference on the Great Recession, Stanford
University, February.
Piketty, Thomas and Emmanuel Saez. 2003. ―Income Inequality in the United States, 1913-
1998.‖ Quarterly Journal of Economics 118 (1) (February): 1-39.
Piketty, Thomas and Emmanuel Saez. 2006. ―The Evolution of Top Incomes: A Historical and
International Perspective.‖ NBER Working Paper No.11955. Cambridge, MA: National
Bureau of Economic Research.
Porter, Eduardo. 2006. ―After Years of Growth, What About Workers‘ Share?‖ New York Times,
October 15.
Poterba, James, Steven F. Venti, and David A. Wise. 2007. ―Rise of 401(k) Plans, Lifetime
Earnings, and Wealth at Retirement.‖ NBER Working Paper No. 13091. Cambridge,
MA: National Bureau of Economic Research. May.
Pryor, Fredric. 2007. ―The Anatomy of Increasing Inequality of US Family Incomes.‖ Journal of
Socio-Economics 36(4): 595-618.
38
Reynolds, Alan. 2007. ―Has U.S. Income Inequality Really Increased?‖ Policy Analysis No. 586.
Washington, DC: Cato Institute. January.
Saez, Emmanuel. 2010. ―Striking It Richer: The Evolution of Top Incomes in the United States
(Updated with 2008 Estimates). Unpublished research note. July.
Schwabish, Jonathan A. 2006. ―Earnings Inequality and High Earners: Changes During and
After the Stock Market Boom of the 1990s.‖ Working Paper No. 2006-06. Washington
D.C.: Congressional Budget Office. April.
Sierminska, Eva and Yelena Takhtamanova. 2006. ―Wealth Effects on Consumption: Cross
Country and Demographic Group Comparisons.‖ Presented at 29th General Conference
of the IARIW, Joensuu, Finland.
Slesnick, Daniel, 1994. ―Aggregate Consumption and Saving in the Postwar United States,‖ The
Review of Economics and Statistics 74 (4): 585-597.
Slesnick, Daniel, 2001. ―Consumption, Needs, and Inequality,‖ International Economic Review,
35 (3): 677-703.
Smeeding, Timothy M. 2005. ―Public Policy, Economic Inequality, and Poverty: The United
States in Comparative Perspective.‖ Social Science Quarterly 86 (5): 955-983.
Smeeding, Timothy and Jeffrey Thompson, 2010. ―Inequality in the Distribution of Income from
Labor and Income from Capital over the Recession,‖ presented at the Tobin Project
Conference on Inequality, April-May 2010.
Tatom, John A. 2007. ―Is Inequality Growing as American Workers Fall Behind?‖ Working
Paper No. 2007-WP-07. Indianapolis: Networks Financial Institute, Indiana State
University. February.
Taussig, Michael K. 1973. ―Alternative Measures of the Distribution of Economic Welfare.‖
Research Report Series: No. 116. Princeton, NJ: Industrial Relations Section, Princeton
University.
U.S. Census Bureau. 2010. ―Income, Poverty, and Health Insurance Coverage in the United
States, 2009.‖ Current Population Reports. P60-238. Washington D.C.: U.S. Census
Bureau.
Walker, Richard. 2005. ―Superstars and Renaissance Men: Specialization, Market Size and the
Income Distribution.‖ CEP Discussion Paper: No. 707. London: Centre for Economic
Performance, London School of Economics and Political Science. November.
Weisbrod, Burton A. and W. Lee Hansen. 1968. ―An Income-Net Worth Approach to Measuring
Economic Welfare.‖ American Economic Review 58 (5) (December): 1315-1329.
Wolff, Edward N. and Ajit Zacharias. 2006a. ―Household Wealth and the Measurement of
Economic Well-Being in the United States.‖ Levy Economics Institute Working Paper
No. 447. Annandale-on-Hudson, NY: Levy Economics Institute, Bard College. May.
39
Wolff, Edward N. and Ajit Zacharias. 2006b. ―Wealth and Economic Inequality: Who‘s at the
Top of the Economic Ladder?‖ Annandale-on-Hudson, NY: Levy Economics Institute,
Bard College. December.
2006-III Share 2009-IV Share
National income $12,093 $12,466
Compensation of employees $7,484 61.9% $7,773 62.4%
Wage and salary accruals $6,075 50.2% $6,266 50.3%
Supplements to wages and salaries $1,409 11.6% $1,507 12.1%
Proprietors' income with inventory valuation and capital
consumption adjustments$1,131 9.4% $1,060 8.5%
Rental income of persons with capital consumption
adjustment$140 1.2% $287 2.3%
Corporate profits with inventory valuation and capital
consumption adjustments$1,655 13.7% $1,468 11.8%
Net interest and miscellaneous payments $662 5.5% $783 6.3%
Taxes on production and imports less subsidies $992 8.2% $1,034 8.3%
Business current transfer payments $84 0.7% $128 1.0%
Current surplus of government enterprises -$5 0.0% -$7 -0.1%
Table 1. Relation of Gross Domestic Product, Gross National Product, and National Income - Including
Those Accounted for in this Paper (shaded)
[Billions of dollars; quarters seasonally adjusted at annual rates]
Source: BEA NIPA Table 1.12, Available at www.bea.gov.
1. We account for supplements to wages and salaries only in so far as they appear as part of defined
contribution pension plans. Health care and other employer subsidies are not counted.
SCF
income
less
capital
plus
finance
plus
retire
plus
home
plus oth
invest
plus
business MCI
SCF to
MCI
as % of
SCF
mean 84,144 73,058 79,292 84,763 92,876 98,868 108,677 110,147 26,003 31%
median (P50) 47,305 43,808 46,157 47,444 51,997 54,488 55,768 55,014 7,709 16%
P90 140,887 128,546 135,571 148,855 163,986 175,709 184,423 185,892 45,005 32%
P95 206,702 185,106 200,588 218,850 241,284 259,486 287,293 290,835 84,133 41%
P10 12,340 11,369 12,340 12,340 13,839 14,397 14,407 14,397 2,057 17%
P99 693,121 516,327 611,309 669,215 728,744 822,229 1,011,830 1,031,528 338,407 49%
90/10 11.4 11.3 11.0 12.1 11.8 12.2 12.8 12.9 1.5 13%
90/50 3.0 2.9 2.9 3.1 3.2 3.2 3.3 3.4 0.4 13%
10/50 0.26 0.26 0.27 0.26 0.27 0.26 0.26 0.26 0.00 0%
95/50 4.4 4.2 4.3 4.6 4.6 4.8 5.2 5.3 0.9 21%
99/50 14.7 11.8 13.2 14.1 14.0 15.1 18.1 18.8 4.1 28%
99/90 4.9 4.0 4.5 4.5 4.4 4.7 5.5 5.5 0.6 13%
gini 0.572 0.539 0.559 0.569 0.562 0.572 0.599 0.608 0.04 6%
Notes:
SCF income
less capital
plus finance
plus retire
plus home
plus oth invest
plus business
MCI
SCF income less income from wealth (interest, dividends, rent, royalties, and income from trusts and non-taxable
investments, including bonds, as well as some self-employment income).
+ imputed flows to stocks, bonds, annuities, and trusts
+ imputed flows to quasi-liquid retirement accounts (401(k), IRA, etc.)
+ imputed flow to primary residence
+ imputed flow to other residences and investment real-estate, transaction accounts, CDs and whole life insurance
+ imputed flow to other assets and businesses + imputed flow to vehicle wealth
- imputed interest flow for remaining debt (after adjusting for negative incomes)
Table 2. SCF (2006-07) - Full Income Definition Summary Statistics - Original Rankings and Long-run Rates of Return
change
Fed default gross household income definition, includes wages, self-employment and business income, taxable and
tax-exempt interest, dividends, realized capital gains, food stamps and other support programs provided by the
government, pension income and withdrawals from retirement accounts, Social Security income, alimony and other
support payments, and miscellaneous sources of income.
SCF
income
less
capital
plus
finance
plus
retire
plus
home
plus oth
invest
plus
business MCI
SCF to
MCI
as % of
SCF
mean 84,144 73,058 79,475 85,181 94,645 100,908 111,131 112,384 28,240 34%
median (P50) 47,305 43,808 46,214 47,602 53,070 55,196 56,858 55,917 8,612 18%
P90 140,887 128,546 135,625 149,259 167,868 179,678 189,333 189,740 48,854 35%
P95 206,702 185,106 200,865 218,977 245,110 265,998 294,841 295,743 89,041 43%
P10 12,340 11,369 12,340 12,340 14,234 14,402 14,503 14,398 2,058 17%
P99 693,121 516,327 613,923 679,215 754,758 842,751 1,040,259 1,062,867 369,746 53%
90/10 11.4 11.3 11.0 12.1 11.8 12.5 13.1 13.2 1.8 15%
90/50 3.0 2.9 2.9 3.1 3.2 3.3 3.3 3.4 0.4 14%
10/50 0.26 0.26 0.27 0.26 0.27 0.26 0.26 0.26 0.00 -1%
95/50 4.4 4.2 4.3 4.6 4.6 4.8 5.2 5.3 0.9 21%
99/50 14.7 11.8 13.3 14.3 14.2 15.3 18.3 19.0 4.4 30%
99/90 4.9 4.0 4.5 4.6 4.5 4.7 5.5 5.6 0.7 14%
gini 0.572 0.540 0.560 0.570 0.563 0.573 0.601 0.610 0.04 7%
Notes:
SCF income
less capital
plus finance
plus retire
plus home
plus oth invest
plus business
MCI
+ imputed flow to primary residence
+ imputed flow to other residences and investment real-estate, transaction accounts, CDs and whole life
insurance
+ imputed flow to other assets and businesses + imputed flow to vehicle wealth
- imputed interest flow for remaining debt (after adjusting for negative incomes)
Table 3. SCF (2006-07) - Full Income Definition Summary Statistics - Original Rankings and Short-run Rates of Return
change
Fed default gross household income definition, includes wages, self-employment and business income, taxable
and tax-exempt interest, dividends, realized capital gains, food stamps and other support programs provided by
the government, pension income and withdrawals from retirement accounts, Social Security income, alimony
and other support payments, and miscellaneous sources of income.
SCF income less income from wealth (interest, dividends, rent, royalties, and income from trusts and non-
taxable investments, including bonds, as well as some self-employment income).
+ imputed flows to stocks, bonds, annuities, and trusts
+ imputed flows to quasi-liquid retirement accounts (401(k), IRA, etc.)
SCF
income
less
capital
plus
finance
plus
retire
plus
home
plus oth
invest
plus
business MCI
SCF to
MCI
as % of
SCF
mean 82,298 71,322 78,891 83,597 88,381 94,396 102,221 104,303 22,005 27%
median (P50) 46,293 43,275 45,027 46,564 49,709 52,499 53,980 53,366 7,072 15%
P90 138,860 129,558 136,123 146,651 157,373 167,196 173,378 175,040 36,179 26%
P95 206,047 181,699 195,087 214,132 224,616 249,409 268,612 272,497 66,451 32%
P10 13,484 12,898 13,042 13,042 14,228 14,592 14,884 14,768 1,284 10%
P99 652,315 489,283 629,802 681,997 708,786 778,926 893,783 934,017 281,702 43%
90/10 10.3 10.0 10.4 11.2 11.1 11.5 11.6 11.9 1.6 15%
90/50 3.0 3.0 3.0 3.1 3.2 3.2 3.2 3.3 0.3 9%
10/50 0.29 0.30 0.29 0.28 0.29 0.28 0.28 0.28 -0.01 -5%
95/50 4.5 4.2 4.3 4.6 4.5 4.8 5.0 5.1 0.7 15%
99/50 14.1 11.3 14.0 14.6 14.3 14.8 16.6 17.5 3.4 24%
99/90 4.7 3.8 4.6 4.7 4.5 4.7 5.2 5.3 0.6 14%
gini 0.561 0.527 0.555 0.565 0.560 0.568 0.590 0.600 0.04 7%
Notes:
SCF income
less capital
plus finance
plus retire
plus home
plus oth invest
plus business
MCI
SCF income less income from wealth (interest, dividends, rent, royalties, and income from trusts and non-taxable
investments, including bonds, as well as some self-employment income).
+ imputed flows to stocks, bonds, annuities, and trusts
+ imputed flows to quasi-liquid retirement accounts (401(k), IRA, etc.)
+ imputed flow to primary residence
+ imputed flow to other residences and investment real-estate, transaction accounts, CDs and whole life insurance
+ imputed flow to other assets and businesses + imputed flow to vehicle wealth
- imputed interest flow for remaining debt (after adjusting for negative incomes)
Table4. SCF (2009 Projection) - Full Income Definition Summary Statistics - Original Rankings and Long-run Rates of Return
change
Fed default gross household income definition, includes wages, self-employment and business income, taxable and
tax-exempt interest, dividends, realized capital gains, food stamps and other support programs provided by the
government, pension income and withdrawals from retirement accounts, Social Security income, alimony and other
support payments, and miscellaneous sources of income.
dpi*
MCI
lesstax
DPI to MCI
lesstax
as % of
DPI dpi*
MCI
lesstax
DPI to MCI
lesstax
as % of
DPI
mean 72,089 105,674 33,584 47% mean 72,089 103,260 31,171 43%
median (P50) 44,409 56,626 12,217 28% median (P50) 44,409 55,511 11,102 25%
P90 120,798 176,554 55,756 46% P90 120,798 171,887 51,089 42%
P95 172,232 277,522 105,290 61% P95 172,232 271,000 98,768 57%
P10 12,372 16,179 3,807 31% P10 12,372 15,962 3,590 29%
P99 520,282 925,758 405,475 78% P99 520,282 910,311 390,029 75%
90/10 9.8 10.9 1.1 12% 90/10 9.8 10.8 1.0 10%
90/50 2.7 3.1 0.4 15% 90/50 2.7 3.1 0.4 14%
10/50 0.28 0.29 0.0 3% 10/50 0.28 0.29 0.0 3%
95/50 3.9 4.9 1.0 26% 95/50 3.9 4.9 1.0 26%
99/50 11.7 16.3 4.6 40% 99/50 11.7 16.4 4.7 40%
99/90 4.3 5.2 0.9 22% 99/90 4.3 5.3 1.0 23%
gini 0.532 0.5806 0.049 9% gini 0.532 0.579 0.047 9%
Notes:
dpi
MCI lesstax
Table 5. After-Tax Concepts (2006-07)
Panel A. Short-run Rates of Return Panel B. Long-run Rates of Return
after-tax concepts change after-tax concepts change
income less federal taxes - calculated with TAXSIM
MCI less federal taxes - calculated with TAXSIM
*Since dpi does not include any imputed flows to wealth, results are the same for short and long term rates of return
(characteristics of household head)
Top 10% Top 5% Top 1% All Bottom 90%
Average age 53.4 55.3 56.6 50.0 49.6
Education Status
Average years of education 15.5 15.8 16.1 13.3 13.0
Share with at least college degree 76.7% 80.2% 87.3% 35.3% 30.2%
Household Status
Share of households headed by
married couple or partners86.9% 86.5% 91.7% 58.8% 55.3%
Share with any kids 46.8% 44.1% 41.5% 43.9% 43.6%
Average # kids (of those with kids) 1.90 1.91 2.04 1.9 1.9
Race
Share non-Hispanic White 86.9% 89.7% 92.0% 70.7% 68.7%
Share Black 2.7% 1.7% 1.8% 12.6% 13.8%
Share Hispanic 2.4% 1.7% 1.5% 9.4% 10.3%
Share "other" 8.0% 6.9% 4.7% 7.3% 7.2%
Working Status
Employed by someone else 53.3% 42.4% 37.7% 59.9% 60.7%
Self-employed or Partner 29.6% 41.3% 45.2% 10.5% 8.1%
Retired/Disabled/Student 15.8% 15.2% 16.2% 25.0% 26.1%
Otherwise not in labor force 1.3% 1.1% 0.9% 4.6% 5.1%
Industry
Agriculture 1.4% 1.4% 0.4% 3.0% 3.2%
Mining & Construction 7.7% 7.1% 12.3% 12.5% 13.2%
Manufacturing & publishing 10.5% 12.0% 9.5% 13.8% 14.3%
Trade, restaurants & bars 11.0% 10.4% 8.6% 15.2% 15.9%
Data, financial, business, repair &
security svcs.16.1% 19.2% 26.1% 12.0% 11.5%
Utility & transport, professional,
scientific, technical, travel, cleaning,
administrative, health, education, &
personal svcs.
46.5% 44.8% 41.5% 36.3% 34.8%
Public admin. & armed svcs. 6.8% 5.1% 1.6% 7.1% 7.1%
Occupations
Executives, managers, scientists,
architects, engineers, lawyers,
teachers, counselors & social
workers, health care practioners,
techs. & support, entertainment,
sports & media
75.7% 83.4% 88.4% 39.1% 33.6%
Technicians, sales, office &
computer operators13.1% 9.7% 9.5% 19.5% 20.3%
Protective svcs., food prep, cleaning
& bldg svcs., personal care, armed
svcs.
4.5% 3.1% 0.0% 11.5% 12.5%
Construction & skilled labor & crafts 4.0% 1.9% 1.0% 18.2% 20.3%
Unskilled labor 2.2% 1.3% 1.1% 10.5% 11.8%
Farm, fishing, forestry, animal
training & care0.5% 0.6% 0.0% 1.3% 1.5%
Table 6. Demographic Profile of Households by MCI Levels - 2007 SCF
0
20000
40000
60000
80000
100000
120000
0
2
4
6
8
10
12
14
16
18
20S
CF
incom
e
less c
apital
plu
s f
inance
plu
s r
etire
plu
s h
om
e
plu
s o
th invest
plu
s b
usin
ess
MC
I
do
llars
(see l
ines)
rati
o (see b
ars
)
Figure 1. Full-income 2006-07 SCF - Long-run returns
90/50 50/10 95/50 99/50 mean median (P50)
Panel A. 99/50 ratios Panel B. 95/50 ratios
Panel C. 90/50 ratios Panel D. 10/50 ratios
Panel E. Means Panel F. Medians
Figure 2. Trend Statistics for Key Income Concepts (long run rates)
0
2
4
6
8
10
12
14
16
18
20
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0
1
2
3
4
5
6
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0
0.5
1
1.5
2
2.5
3
3.5
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0
10000
20000
30000
40000
50000
60000
70000
80000
90000
100000
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0
10000
20000
30000
40000
50000
60000
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04
SCF income plus finance plus home MCI
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
0.0
1.0
2.0
3.0
4.0
5.0
6.0
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
$0
$20,000
$40,000
$60,000
$80,000
$100,000
$120,000
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
$0
$10,000
$20,000
$30,000
$40,000
$50,000
$60,000
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
SCF income plus finance plus home MCI
Panel A. Growth between 1989 and 2007 by percentile of SCF and MCI distribution (long-run rates)
Panel B. Difference in MCI and SCF Growth Rates by percentile
Figure 3. MCI and SCF Growth Compared
20%
40%
60%
80%
100%
120%
140%
160%
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
perc
ent change 1
989-2
007
percentile
SCF
MCI
-10%
0%
10%
20%
30%
40%
50%
60%
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
gap in
gro
wth
rate
s
percentile
MCI less SCF
(with slopes)
Panel A. Top one percent
Panel B. 95 to 99 ptile
Panel C. 90 to 95 ptile
NOTES:
1. MCI is based on long-run rates of return.
2. CBO uses a measure of "comprehensive income" that includes realized capital gains.
3. WZ is "wealth-adjusted" income from Wolff and Zacharias, May 2006.
4. PS figures include capital gains income.
Figure 4. Comparing Income Shares of Top Fractiles (1989-2009)
8
10
12
14
16
18
20
22
24
26
1989 1992 1995 1998 2001 2004 2007 2008 2009proj
Inco
me s
hare
Slopes1989-2007: MCI: .95 CBO: 1.04 PS: 1.43
1989-2001: MCI: 1.12 CBO: .78 WZ: .85 PS: 1.19
8
10
12
14
16
18
20
22
1989 1992 1995 1998 2001 2004 2007 2008 2009proj
Inco
me s
hare
MCI WZ PS CBO
Slopes1989-2007: MCI: .47 CBO: .11 PS: .18
1989-2001: MCI: .43 CBO: .12 WZ: .33 PS: .28
6
8
10
12
14
16
18
1989 1992 1995 1998 2001 2004 2007 2008 2009proj
Inco
me s
hare
Slopes1989-2007: MCI: -.09 CBO: -.06 PS: -.1
1989-2001: MCI: -.22 CBO: -.01 WZ: -.1 PS: -.06
Panel A. Gini Indices
Panel B. 99/50 Ratios
Panel C. Income Share of Top 5%
Figure 5. The Impact of Including Vehicle Service Flow on Inequality Trends
.44
.46
.48
.50
.52
.54
.56
.58
.60
.62
.64
1989 1992 1995 1998 2001 2004 2007 2009_proj
SCF
MCI - baseline
MCI - vsf
4
6
8
10
12
14
16
18
20
1989 1992 1995 1998 2001 2004 2007 2009_proj
MCI - baseline
MCI - vsf
SCF
22%
24%
26%
28%
30%
32%
34%
36%
38%
40%
42%
1989 1992 1995 1998 2001 2004 2007 2009
MCI - Baseline
MCI - VSF
SCF
Figure 2. Labor and Capitol Shares (MCI Long-run Rates)
Figure 6. Labor and Capital Shares - SCF and MCI Gross Income
Panel A. SCF Gross Income
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009PROJ
perc
en
t
Labor1 Capital2 SEI3 Other4
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009PROJ
perc
en
t
Labor5 Capital5 Other
Smoothed 3 percentile averages using long-run rates
Panel A. Labor, Capital, and Other Share of MCI by percentile - 2007
Panel B. Labor and Capital Shares of MCI by percentile - 1989 and 2007
Figure 7. Labor and Capital Shares of MCI by percentiles of MCI distribution
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
5 8 11 14 17 20 23 26 29 32 35 38 41 44 47 50 53 56 59 62 65 68 71 74 77 80 83 86 89 92 95 98
Sh
are
of M
CI
percentile of MCI distribution
other% '07
cap% '07
lab% '07
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
5 8 11 14 17 20 23 26 29 32 35 38 41 44 47 50 53 56 59 62 65 68 71 74 77 80 83 86 89 92 95 98
Sh
are
s o
f M
CI
percentile of MCI distribution
lab% '89
cap% '89
lab% '07
cap% '07
Panel A. SCF Income
Panel B. MCI
Figure 8. Labor Share of Income of High-Income Households, by Income Concept
y = 0.0059x + 0.4238
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1989 1992 1995 1998 2001 2004 2007 2009_proj
Lab
or
Shar
e o
f SC
F In
com
e
TOP 1%95-99PTILE90-95PTILELinear (TOP 1%)
y = -0.0144x + 0.4559
0.3
0.35
0.4
0.45
0.5
0.55
0.6
0.65
0.7
0.75
1989 1992 1995 1998 2001 2004 2007 2009_proj
Lab
or
Shar
e o
f M
CI
Type of Income def
return
categories
Long-
Run
rates 1977-07 1987-89 1990-92 1993-95 1996-98 1999-01 2002-04 2005-07 notes
SCF income Fed gross income
less capital SCF income less income
from wealth (rent, interest,
dividends,
trusts&annuities)
SI 0.07 0.095 0.147 0.070 0.152 0.210 0.044 0.036 0.073
SIBI* 0.06 0.085 0.116 0.074 0.108 0.135 0.050 0.039 0.059
BI 0.05 0.076 0.086 0.078 0.065 0.060 0.055 0.043 0.045
CPI 0.03 0.042 0.043 0.040 0.026 0.021 0.025 0.026 0.035
plus retire + imputed flows to quasi-
liquid retirement accounts SI 0.07 0.095 0.147 0.070 0.152 0.210 0.044 0.036 0.073
plus home + imputed flow to primary
residence HI 0.06 0.059 0.060 0.023 0.025 0.041 0.064 0.074 0.070
SI 0.07 0.095 0.147 0.070 0.152 0.210 0.044 0.036 0.073
CPI+1 0.04 0.052 0.053 0.050 0.036 0.031 0.035 0.036 0.045
BI 0.05 0.076 0.086 0.078 0.065 0.060 0.055 0.043 0.045
BICPI* 0.04 0.059 0.064 0.059 0.046 0.041 0.040 0.034 0.040
plus business + imputed flow to other
assets and businesses +
imputed flow to vehicle
wealth
SI 0.07 0.095 0.147 0.070 0.152 0.210 0.044 0.036 0.073
MCI - imputed interest flow for
remaining debts (after
replacing finc5 with SCF
income when
finc5<SCFincome)
CPI + 6 0.09 0.102 0.103 0.100 0.086 0.081 0.085 0.086 0.095
Appendix Table 1. Rates of Return Applied to Different Portions of Full-Income
plus finance + imputed flows to stocks
and bonds + imputed
flows to annuities and
trusts
* Average of SI and
BI is for "combination"
mutual funds, CPI is
for tax-free bonds
plus oth invest + imputed flow to other
residences and investment
real-estate + imputed flow
to transaction accounts +
imputed flow to CDs and
whole life insurance
*Whole life insurance
is given BI rate, CDs
are given average of
BI and CPI
55
Housing index
(HI)
Stock Indices
(SI)
Bond indices
(BI)
Inflation
(CPI)
A. "Short Run"
1989 6.0% 14.7% 8.6% 4.3%
1992 2.3% 7.0% 7.8% 4.0%
1995 2.5% 15.2% 6.5% 2.6%
1998 4.1% 21.0% 6.0% 2.1%
2001 6.4% 4.4% 5.5% 2.5%
2004 7.4% 3.6% 4.3% 2.6%
2007 7.0% 7.3% 4.5% 3.5%
B. "Long-run"* 6.0% 7.0% 5.0% 3.0%
*Rates used for 1988-89 to 2006-07
Appendix Table 2. Short Run (three-year average) and Long Run (1988-
2007) Rates of Return
56
Income
Matching Source. Table (Row
Number) Source Detail
Percent Change
2007 Q3/4 to 2009
Q3/4 change
Interest NIPA. 2.1 (14) -5.8%
Dividends NIPA. 2.1 (15) -28.6%
Non-taxable Investment Income NIPA. 2.1 (14) *SCF detail refers to bonds* -5.8%
Other business/investment/rent/trust NIPA. 1.12 (9, 39) Combined rental and proprietor 5.7%
Earnings Analysis of CPS ORG, Jan. to Nov. varies by industry,
education
Proprietors income NIPA. 2.1(9) -4.4%
Capital gains CBO Jan. 2009 Budget Outlook Anticipated tax revenue decline of 40% -40.0%
Public Transfers (excluding Soc. Sec.) NIPA. 2.1(17 less 18) 36.2%Retirement Income (including Soc.
Sec.)NIPA. 2.1(18)
15.3%
Assets
CDs FOF. B.100(12) time and savings deposits 4.9%
Stocks FOF. B.100(24) corporate equities -21.6%
Stock mutual funds FOF. B.100(25) mutual fund shares -12.6%
Bonds FOF. B.100(18) treasury securities 404.2%
Other bond mutual funds FOF. B.100(21) corporate and foreign bonds 21.9%
Savings bonds FOF. B.100(17) savings bonds -2.5%
Govt. Bond Mutual Funds FOF. B.100(19) agency and GSE-backed securities -83.7%
Tax-free bond mutual funds FOF. B.100(20) municipal securities 9.2%
Combination and other mutual funds FOF. B.100(25) mutual fund shares -12.6%
Other (trusts, annuities, etc.) FOF. B.100(30) miscellaneous 10.8%
Home equity FOF. B.100(49) owner's equity in household real estate -41.0%
Quasi-l iquid retirement Urban Insti tute Analys is of FOF www.urban.org/retirement_policy/url.cfm?ID=411976 -14.0%
Transaction accounts FOF. B.100(11) (checkable deposits) 140.1%
Life Insurance FOF. B.100(27) life insurance reserves asset 3.8%
Nonresidential real estate FOF. B.100(49) owner's equity in household real estate -41.0%
Other residential real estate FOF. B.100(4) modify in same way as residential real estate -21.4%
Debt for other residential property FOF. B.100(33) home mortgages -1.3%
Other financial assets FOF. B.100(30) miscellaneous assets 10.8%
Other nonfinancial assets FOF. B.100(7) and (30) combined consumer durables or miscellaneous assets 9.8%Business with active or nonactive hh
interestFOF. B.100(29) equity in non-corporate bus.
-23.6%
Vehicles FOF. B.100(7) consumer durables or miscellaneous assets 9.6%
Total debt FOF. B.100(31) total l iabil ities -1.4%
Mortgages and home equity loans FOF. B.100(33) home mortgages -1.3%
Home equity l ines of credit FOF. B.100(33) home mortgages -1.3%
Appendix Table 2a. Adjustments made to SCF Income and Asset Categories for 2009 Projection
57
Dropout HS Only
Some
College, No
Degree
Bachelor's
Degree or
More
Agriculture, Fish, Forest, Construction -23.9% -13.1% -10.1% 4.9%
Manufacturing, Information, Publishing -21.4% -12.7% -9.3% 1.2%
Trade, Leisure, Restaurants -9.1% -3.4% -1.1% 2.2%
Utilities, Professional, Educational, Health Services -10.0% 2.0% 7.1% 6.9%
Data, Finance, Other Services -6.1% -0.8% -2.5% 2.7%
Public Administration -22.0% 3.4% 3.8% 9.4%
Appendix Table 2b. Change in Earnings between 2007 and 2009, by Education and Industry
Note: Earnings change figures estimated from Current Population Survey, Outgoing Rotation Group data, for January through
November of 2007 and 2009. The education and industry groupings are based on the categories in the public SCF data. Earnings
change is the difference between the cummulative weekly earnings for each industry/education cell in 2007 and 2009. Differencing
total earnings reflects changes in employment, weekly hours worked, and wages.
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
99/10 ratio 50.5 40.9 45.3 54.2 71.7 63.0 71.6 63.2
99/50 ratio 11.5 10.2 10.4 13.5 17.4 15.6 18.8 17.5
10/50 ratio 0.23 0.25 0.23 0.25 0.24 0.25 0.26 0.28
gini 0.560 0.526 0.540 0.561 0.595 0.579 0.608 0.600
1988-89 1991-92 1994-95 1997-98 2000-01 2003-04 2006-07 2009_proj
mean 33.9% 35.8% 35.5% 38.0% 40.0% 45.2% 50.8% 40.9%
median 20.0% 21.9% 17.2% 19.6% 22.4% 21.9% 25.6% 20.8%
99th ptile 65.2% 58.1% 68.9% 77.0% 90.4% 104.2% 99.8% 78.3%
99/10 ratio 32.7% 29.9% 37.1% 41.2% 63.6% 69.7% 57.8% 59.1%
99/50 ratio 37.6% 29.7% 44.1% 48.0% 55.6% 67.5% 59.1% 47.6%
10/50 ratio 3.6% -0.2% 4.9% 4.6% -5.1% -1.3% 0.8% -7.2%
Panel A. MCI1
Appendix Table 3. Basic Trends from SCF - Comparisons over time - Long run rates
1 MCI (more complete income) subtracts capital income (except realized capital gains) from Gross Income
and adds back f low s to assets and debt.
2 SCF net some capital income takes Gross Income and subtracts interest, rent, dividends, and annuity
and trust income, but retains realized capital gains.
For details on the definitions and rates used in developing Full Income see Tables 3 and 4.
Notes:
Panel B. % change bet SCF net some capital income to MCI2
58
by networth by SCF income by MCI
Addendum:
MCI by percentile
(short-run rates)
p10 11 68,039 23,112 25,664
p50 117,033 168,848 152,491 135,278
p90 870,988 876,835 864,138 761,991
p95 1,686,125 1,491,843 1,645,577 1,488,262
p99 6,252,244 5,607,287 6,509,146 6,443,411
Appendix Table 4. Values of networth by alternative rankings - 2007